V6 · Trader

1. What this module is, and why it exists

This volume is the trader seat’s home. It starts at Trader Pools (§2), opens the book behind one pool, and walks Rules, Forecast and Outcomes under Allocation, the Desk, nominations, BESS and the Dashboard. Off-taker screens live in V5 · OffTaker. IPP screens live in V4 · IPP. The shared three-seat picture stays in V3 · Strategy.

Creating a pool (licence, rights, treatments) is §2B. The pool as an entity stays in V2 §5. This volume then opens the book when you need to trade.

1.1 The trader's problem

An energy trader in the South African wheeling market sits between generators and consumers:

• You buy energy from IPPs (independent power producers — solar farms, wind farms, firm/dispatchable blocks) under PPAs, at contracted c/kWh prices.
• You sell that energy to off-takers (factories, malls, municipalities) and to other traders, under sell-side PPAs, at higher c/kWh prices.
• You pay wheeling charges to move the energy across the grid, and you lose a few percent of every MWh to distribution losses on the way.
• Your profit is the spread: sell price − buy price − wheeling, hour by hour.

The catch is that none of this is flat. Solar produces at midday and nothing at night. Your factory load runs 24/7; your mall load peaks in the daytime; a firm evening block must be served no matter what. Time-of-use tariffs make an evening kWh worth far more than a midday one. So the questions that decide whether you make money are inherently hourly:

• Which supply serves which demand, in each of the ~8,760 hours of the year?
• How much supply goes unsold (spill — cost with no revenue)?
• How much demand goes unmet (shortfall — expensive grid top-ups or penalties)?
• What margin does each hour actually earn, and how much of it evaporates if generation comes in below forecast?
• How much flexible energy should you nominate for next month, when nominating too little means expensive grid top-ups and too much means paid-for energy that spills?
• If you own a battery, when should it charge and when should it discharge, given that every stored kWh loses a little to efficiency and wears the warranty?

Doing this on a spreadsheet is where portfolios go to die. This module does it with a deterministic engine and puts the answers on eight screens.

1.2 What the module does

You group your contracts and sites into pools (supply on one side, demand on the other), and the engine allocates every forecast hour of energy from supply to demand, pricing each flow with the same conventions the settlement system uses. Every result is stored as tagged, auditable allocation lines — so coverage, margin, spill and shortfall are never estimates; they are sums over hourly lines you can drill into. On top of the engine sit the decision screens summarised below.

Auditable allocation lines

An allocation line is one immutable row for one SAST hour: which member of which pool supplied which member of which pool, how much energy was sent and delivered, the buy / sell / wheeling legs in c/kWh, the Rand margin, and which rules produced it. Headline KPIs in Performance and the Dashboard are simply sums (or ratios) over those rows — open a run's Allocations tab (§8) and you can take any number apart.

Four common kinds (illustrative values, same conventions as settlement):

Because every line is tagged, "coverage is 94%" means these hours and members were covered; "spill cost R 2 160" means this exact unsold row. Re-running creates a new run; old lines stay for audit (§1.3).

Decision screens on top of the engine

• Simulation workbench (§9) — safe what-if: run the book as today (baseline) and again with a pretend buy/sell candidate and terms (variant). Compare deltas (margin, spill, shortfall, concentration) without touching live membership, forecasts or nominations; promote only when you like the story.
• Nomination workbench (§10) — for flexible supply members, propose an hour-by-hour monthly declaration using asymmetric under-/over-nomination costs (newsvendor / critical quantile), show the full rationale, let you override any hour, then submit into settlement only after you approve.
• BESS dispatch view (§11) — hour-by-hour charge/discharge schedule against the book's real spill and shortfall, subject to SoC, power, efficiency, cycle-cap and minimum spread; every move (and every profitable-looking pair it declined) is recorded with a rationale.
• Risk dashboard (§12) — drillable panels for portfolio margin, risk (P50 vs P90), prices, concentration, limits/alerts, battery tiles and nomination performance — all refreshed from the latest snapshot so you diagnose what and where, not just that something is wrong.

Off-taker and IPP seats reuse the same engine with buy- or sell-shaped screens (§12–§14); the trader path above is the fullest tour of what "Strategy" means.

1.3 Design principles you will feel in the UI

A run is one complete execution of the allocation engine: given a scenario, a calendar month and a forecast grade (P50 or P90), the engine walks every SAST hour, matches supply to demand under the book's rules, and writes the resulting allocation lines (§1.2). The run is then frozen — coverage, margin, spill and shortfall are simply sums over those lines. You create runs from the Allocation tab’s Forecast mode (§6), open any one in §8 to inspect the proof, and treat the newest non-superseded COMPLETE run for a scenario / grade / period as the current answer for that case. Four deliberate decisions shape how runs (and the screens around them) behave:

1. Runs are immutable. Once an allocation run completes, its lines never change. Re-running the same scenario and period creates a new run and marks the old one superseded (kept, greyed out). Why: audit trail. You can always answer "what did the book say last Tuesday?", and a settlement dispute can be defended with the exact run that informed a decision.

2. Every number is a sum over tagged lines. Each line records which member of which pool supplied which member of which pool, in which hour, at which prices, and which rules produced it. Why: when a KPI looks wrong, you drill down to the exact PPA pair and hour instead of arguing about a rollup. The same line-by-line convention reconciles with settlement.

3. Every run pins its inputs with a hash. The run header shows an inputs hash — a SHA-256 fingerprint of everything the run consumed (members, prices, rules, forecasts, period — and the battery's dispatch parameters). Same hash ⇒ byte-identical results. Why: reproducibility. "Why did run X say margin was R 100/MWh?" — "Here is its hash; re-run with those inputs and you get the same lines."

4. Scenarios never mix. Every run is labelled by the kind of input it consumed — not by a cosmetic tag. The four labels mean:

   Label	What the run does	Why it is labelled that way
   FORECAST	Allocates the book's rolling supply-forecast shapes against demand profiles for the chosen month and grade	It is the operating plan — "what we expect to happen" — and is the default current position in Performance
   NOMINATED	Re-runs the engine using the submitted nomination rows as the declared volumes, not the free forecast	It answers "what did we say we would deliver?" so you can compare declared position to the forecast plan
   ACTUAL	Allocates using settled meter data for the period (when that data exists)	It is labelled for realised performance — outcome review after the fact, not forward planning
   SIMULATION	Allocates against hypothetical inputs (a candidate PPA, a tweaked shape, an isolated what-if)	The name marks it as non-live: it never supersedes your forecast or nominated current, so a sandbox case cannot contaminate the book's real position

   A simulation never replaces your forecast; a later nominated run does not overwrite the forecast current simply because it is newer. Filter and compare within one scenario (§6.3).

   The operating-plan forecast itself has an explicit source order. A human :MANUAL upload still wins; a weather-driven :MODEL series sits below that and above the synthesised per-technology shape. Hours the model covers take dated values; hours beyond the horizon keep the shape. The operator-facing account of where those model numbers come from lives in V4 §17.

The same philosophy extends to nomination and batteries: every nomination the optimiser proposes carries its full rationale (the scenarios it considered, the asymmetric costs, the quantile it aimed for, and what the naive alternatives would have cost), and every hour the battery moves — and every profitable-looking hour it deliberately did not — is recorded and shown. You are never asked to trust a number you cannot take apart.

1.4 Concepts to have in hand before you start

2. Trader pools — the directory

You need to find the pool that needs action before opening its work. A trader pool groups the rights, contracts and protections under which your organisation buys and sells energy. One organisation can operate several pools. You work on one pool at a time.

Open Wheeling → Trader Pools (/org/trader-pools). Creating or editing the registration record is taught in §2B.

Follow along. Seat: trader_1@webill.net (password supersecret). Path: Trader Pools → T1 Renewables Pool. You should see the directory table, then the pool’s Operations view with five workspace tabs.

2.1 The directory

The financial verdict sits above search. Each table row shows the pool name, contracts and members, lifecycle status, coverage and next step. A pool with a forecast also shows its margin, spill, delivery window, forecast target date, generation time and source allocation run. A future target date is a forecast date; Generated tells you when the record was produced.

Attention first ranks a forecast loss ahead of high spill, other forecast issues, missing members and missing forecasts. A risk reason links to its repair: Commercial for unsold supply, Allocation for missing members, Performance for a forecast, or Operations for a missing SAWEM trading unit. Needs attention filters these rows; search, status and comparison choices survive a visit to the workspace and a return through Pools.

Not assessed · no members means there is nothing to forecast. Not assessed · awaiting forecast means members exist but no snapshot is available. These rows show — for coverage and financial rates. A measured zero appears only when there is an assessment. Active describes the registration lifecycle; it does not mean the pool earns money or is ready for every market capability.

Filters & columns opens status, sorting and optional margin, risk and PPA columns. Selecting a PPA count opens the matching Deals register. Portfolio summary & coverage notes expands the fleet metrics. How to read this links to this manual.

A missing SAWEM (South African Wholesale Electricity Market) trading unit concerns that pool’s market participation. It does not prevent PPA pool creation. Organisation accreditation, pool lifecycle and market registration are separate checks. If the organisation cannot trade, complete its accreditation in V2 §2.

2.2 The detail drawer

Select a pool name to open Operations. Pool details opens the branded right-hand drawer with the operating record, regulator preview, available Trading Units, verdict and Edit & complete pool record. Close it with the X, Escape or backdrop. Archive asks for confirmation and marks a pool Suspended; Activate restores an eligible non-active pool.

2.3 Getting in — the trader strategy area

flowchart LR
  A[Trader Pools directory] --> B[Select a pool name]
  B --> C[Operations]
  C --> D[Choose a workspace tab]
  D --> E[Review evidence and act]
  E --> A

1. Open Trader Pools and find the pool by name.
2. Select its name to open Operations at /org/trader-pools/pools/:id/operations.
3. Choose the workspace tab for the task. Check the pool name and period first.
4. Use Pools to return with your directory search and comparison choices.

Your pool’s strategy book owns forecasts, allocations and risk records. It is created automatically when needed. An unavailable or foreign pool returns you to the directory. Legacy /org/trader-pools/:id/strategy links still enter the corresponding workspace task.

2.4 The strategy tabs

Older links and chapter anchors retain the names Overview (now Performance), Book (Allocation) and Desk or Simulations (Commercial). A book remains the underlying portfolio record; it is not a separate workspace tab.

Commercial shows a calendar delivery month; Performance shows the latest snapshot’s allocation-run window. They may differ. Read the visible dates and model basis before comparing energy or margin. The Performance headline is the deterministic snapshot result; its separate Monte-Carlo fan shows a distribution and a median, which can produce a different number.

View pool contracts opens an owned, pool-filtered register. Open a contract, then Back to Contracts to retain the pool and register filters. Back to pool returns to the originating workspace and delivery month.

Where this is taught. Registration drafts and applying edits: §2B. Market Trading Units: V2 §5.

2B. Onboarding a trader pool — the licence file

A factory or a wind plant is a place with a meter. A trader pool is not. It is a licensed trader’s trading book: a bundle of rights, treatments and protections under which you buy from generators and sell to consumers. Until that book exists on this platform, there is no directory card, no Strategy book, and nothing to forecast.

The papers do not arrive on one afternoon. If you had to type everything in one sitting, you would lose the work the first time you closed the tab.

Think of a new shop joining the merchant’s list. You start a licence file on the first visit and add stamps as the papers arrive. You do not open the shop until the core stamps are there. That file is the trader pool draft. The live pool is created only when you press Create trader pool.

This is not standing up a generator plant. A plant has a pin, a meter and a technology. This workspace is the pool’s licence, operating and commercial file.

Where this is taught. Generator-site onboarding (site, meter, technology, connection): V2 §4 · V4 §3. Off-taker site onboarding: V5 §2.3. The pool as an entity, accreditation and market Trading Units: V2 §5.

The create path is not a five-step modal. Add New on Trader Pools writes a draft and opens a full-page workspace at /org/trader-pools/onboarding/<case id>. Edit on an existing card opens the same workspace, prefilled from that pool — not a blank draft, and not a second book.

Follow along. Seat: trader_1@webill.net (password supersecret). Path: Trader Pools → the round + (Add New Pool, #open-trader-pool-wizard). You should land on Profile & licence with title Untitled trader pool and a Draft chip. When you have finished reading, press Discard so this seed does not keep a leftover card.

Data. full_demo · seed 42 · Trader 1 · a clean seed has no drafts · last checked 2026-09-18.

2B.1 The trader’s problem

You need a named book that counterparties and the regulator can recognise: who you are, which wires you may use, what you may do, who is accountable, how charges land, and what you promise buyers. Typing that into a one-shot wizard fails the first time a licence date or a policy reference is still in the post.

The workspace lets you pause. Each section autosaves. You return from the amber Trader pool drafts rail on /org/trader-pools until the six core sections pass.

2B.2 How you get here

You must already sit in a trader seat — an organisation accredited to trade. If it is not, Trader Pools shows an amber notice instead of the create button: “The Organization has not been accredited to trade either by Eskom or by any municipality which supports wheeling.” Fix V2 §2 first.

The page heading is Trader pool onboarding on a draft and Edit trader pool on a live pool. The eyebrow is Draft trader pool or Registered trader pool. ← Trader Pools returns to /org/trader-pools without discarding.

A foreign or unknown case id never opens. You bounce to the directory with “That onboarding draft is unavailable for this organization.”

2B.3 Words on this workspace

The six core sections that must pass before create are Profile & licence, Network & roles, Activities & limits, Operations & data, Commercial & risk, and Customer protections. Market standing data never blocks create. A ready draft usually reads 6 of 7.

2B.4 How to read the checklist badges

The progress line reads N of 7 sections complete. Beside it the save state says Draft created — no answers saved yet, Draft saved just now, or Last saved {day} at {time}.

The readiness tile is amber while core sections are missing (Draft is safely stored · N core sections still need information). It turns green when create is allowed (Ready to create pool · Core registration controls are complete). In edit mode it always says Draft changes are pending.

Check & mark complete confirms a complete section or shows inline missing-field feedback. Required fields have a star; required licence fields appear when that licence posture is selected. Draft saved does not mean section complete. Review readiness is the same reading in one place. Next: {section} beside it moves to the next section; on the last one, Market standing data, it reads Review readiness.

The section header counts the core sections on their own — Core section 2 of 6, and Optional section 1 of 1 on Market standing data — and stays pinned at the top while you scroll a long section. On a window narrower than 1280 px the checklist sits above the form as a compact two-column grid; picking a section there scrolls the form into view.

2B.5 Explanations drawer

Explanations (#trader-pool-onboarding-explanations) opens a right-hand list for this section (#trader-pool-onboarding-explanations-drawer). Think of it as the card catalogue for the form: find the word, click a choice if you want it written in, close the drawer, keep working.

The header is branded green. The title is Explanations. The eyebrow repeats the section name. Close with the header X (#trader-pool-onboarding-explanations-close), press Explanations again, or press Escape. The arrow beside Explanations points left while the drawer is open.

On a window 1280 px or wider the page makes room and the drawer sits beside the form, so the form stays usable; the checklist shrinks to an icon rail (#trader-pool-onboarding-section-rail), one icon per section with its name on hover. Narrower, the drawer would cover the form, so it opens over a dimmed backdrop (#trader-pool-onboarding-explanations-backdrop), which also closes it. Its intro says which you have — You can keep filling the form while this panel is open. or Close the panel to return to the form.

Most fields here have no ⓘ on the label. Compulsory inputs still show the red *. Topics with choices (status, licence posture, visibility, network rights, roles, activities, treatments, protections) fill the matching input when you click them. Topics without choices (name, emails, document references) are notes only — you still type those.

The drawer stays open when you jump sections, and the topics change to match. NERSA licence topics appear only after you pick NERSA trading licence required. Cross-border topics appear only after Permitted.

2B.6 Create or edit — the chrome

1. Open Trader Pools. Confirm you may trade (the + is present).
2. Press Add New to start a draft, or open a card and press Edit to reopen a live pool.
3. Work the Pool checklist. Jump with the checklist, or press Next: {section} at the foot of each section. Each change autosaves. Use Explanations when a word is unfamiliar.
4. Press Check & mark complete on a section when you think it is done. Inline feedback lists what is missing and focuses the first missing field.
5. Keep draft & exit (#trader-pool-onboarding-save-exit) returns you to /org/trader-pools without creating or updating the pool. Draft answers stay on the amber rail. Edit answers stay on the edit case so you can reopen them from Pool details. The confirmation explicitly says the live pool is unchanged.
6. Review readiness on a new draft, or Review & apply changes on an edit, (#trader-pool-onboarding-review) opens the modal. Every section badge is listed, with repair links for missing information. Tab stays inside the review; Escape closes it and returns focus to the review button.
7. On a draft, Create trader pool (#trader-pool-onboarding-create-pool) is shown only when nothing core is missing. On success the case is ACTIVATED, the live pool is written, and you return with {name} was created successfully.
8. On edit, the same button reads Apply changes to live pool. It writes the answers onto the existing pool and returns with {name} was updated successfully. It does not create a second book.
9. Discard (#trader-pool-onboarding-discard) exists only on a draft. It asks first, then deletes the case and never creates a pool. Edit has no Discard — you cannot throw the live pool away from here.

Locked on edit (#trader-pool-onboarding-roles-locked): Eligible supply roles, Eligible demand roles, and Nomination capability. They are set at creation and stay as registered. The notice on Network says so. Save leaves those three alone even if the form is tampered with.

Dropdowns on this workspace are listed A–Z by label (Active before Suspended; Private before Public; Eskom and municipality before Eskom only).

2B.7 Profile & licence

Section hint on screen: “Pool identity, licence or exemption evidence and effective dates.”

Who this shop is, and whether it may trade under a full NERSA licence or a Schedule 2 exemption. An amber Licence evidence note is always shown: Eskom virtual wheeling still wants the evidence reference.

Defaults on a new draft: Schedule 2 exemption / registration applies, Active, Public.

Empty Pool name or missing evidence keeps Profile at Needs evidence. Picking Suspended keeps create blocked even if every other field is filled.

2B.8 Network & roles

Section hint: “Access mandate, trading roles and cross-border authority.”

Which wires this book may use, which generator and buyer shapes it may take, and whether it may cross a national border. This is the pool’s mandate, not a site connection.

Cards you are not accredited for do not appear. A trader with only Eskom accreditation sees Eskom only. Both accreditations unlock Eskom and municipality. The default is the first of Eskom · Municipality · Both that you hold.

Defaults on a new draft: cross-border Not permitted, supply Flexible, demand Trader sales, nomination Nomination capable.

A plant you later add must share at least one supply class with this list. That matching is taught with generator-site classes in V2 §4.2.

Careful. Cross-border Permitted without both references flashes Still needed: import/export licence, SAPP portfolio.

2B.9 Activities & limits

Section hint: “Permitted scope, restrictions and compliance controls.”

What the licence allows, and what it forbids. Defaults: Buy energy, Sell energy, Wheel energy ticked; restrictions empty.

2B.10 Operations & data

Section hint: “Accountable contacts, secure interfaces and reconciliation.”

Who runs the book day to day, and whether the secure link that carries nominations, meter data and settlement files is ready. A blue Operational accountability note is always shown: SAWEM wants that link; NERSA’s draft bilateral rules want a monthly energy and money reconciliation path.

Default: Secure data-interface readiness is Not assessed. Create will not accept that default — the interface must be Ready.

2B.11 Commercial & risk

Section hint: “Procurement targets, wheeling charges and balancing exposure.”

How you aim to buy, how network charges reach customers, and who carries imbalance. The megawatt targets are planning figures. They do not create a deal.

Defaults: wheeling Full pass-through, balancing Trader-borne.

Where this is taught. How settlement applies wheeling and imbalance: V8 §7.

2B.12 Customer protections

Section hint: “Safeguards, complaints, consent and dispute procedures.”

What this pool promises the buyers it serves. Defaults already ticked: Price caps, No unilateral price changes, Billing transparency, Balancing-cost protection, Dispute resolution. You may add or remove cards; at least one must remain.

2B.13 Market standing data

Section hint: “SAWEM Party, agreements, accounts and credit cover remain authoritative.”

This section is a placeholder handoff, not a form. There are no inputs and no Check & mark complete; as the last section it ends with Review readiness. Market Participant identity, MPID, Market Participation and Balancing Agreements, settlement accounts and credit-cover providers belong to the organisation-level Market registration. This pool links to those records instead of copying them.

The card heading is Keep SAWEM standing data authoritative. Open Market registration goes to /org/brp/registry.

You can still create the pool and trade on a physical or virtual PPA without SAWEM registration. The pool is then not yet a market participant. The badge stays Optional.

Where this is taught. Trading-unit registration: V9 §3.9B. What a Trading Unit is: V9 §3.3.

2B.14 What “done” looks like

1. Review readiness lists every badge. Missing core titles appear as a comma-separated list (Profile & licence, Operations & data).
2. Create trader pool is shown only when nothing core is missing. Create on an incomplete draft stays hidden; the modal title is More evidence is needed.
3. On success the draft is ACTIVATED, the live pool is written with the answers (and any onboarding-evidence references), and you return to the directory with {name} was created successfully.
4. Done on this seat is: a quiet card with the name you typed, status Active, and no amber draft row for that id. Strategy still needs a first Performance click (§2.3 · §3).

Keep draft & exit on a draft leaves an amber rail Trader pool drafts (#trader-pool-onboarding-drafts) above the pool cards. A clean Trader 1 seed has no drafts until you press +.

Where this is taught. The directory that receives the new card: §2. Opening the book: §3. Wiring members: §5. Generator plants you may later buy from: V2 §4. Off-taker shops you may later sell to: V5 §2.3.

3. Getting in — open the book behind a trader pool

Your commercial portfolio can look healthy in the directory while still having no forecast, allocation or risk view. Strategy gives each trader pool a book: the system record that turns its contracts into a plan you can inspect and act on.

Data. full_demo · seed 42 · stage S09 (settlement matrix) · last re-derived 2026-07-18 against the full_demo manifest.

3.1 Pool, book and seat

A Trader Pool is the commercial grouping you manage: its generators, off-takers, prices, rules and coverage targets. A Trader Book is created for that pool when you first open Strategy. It owns its runs, allocations, simulations, nominations, battery schedules, snapshots and limits.

The seat determines the question you ask. A trader seat balances supply against demand; an off-taker seat starts with savings; an IPP seat starts with revenue and unsold energy. They use the same engine but are entered from their own directories. §12 maps those other seats without changing the trader workflow taught here.

You never create the book in a separate setup wizard. Opening a pool workspace creates it if needed. Select Performance to inspect its forecast. That first visit can legitimately have no snapshot yet: it means no forecast has been run, not that the book is broken.

3.2 Open Strategy for the right pool

Start at Wheeling → Trader Pools (/org/trader-pools). Select the pool name to open Operations, then select Performance, Allocation, Commercial or Dashboard. The pool workspace uses /org/trader-pools/pools/:id/:section.

Check the pool name in the header: every number belongs to that pool. Unknown or foreign pool IDs return you to the directory.

Follow along. Seat: trader_1@webill.net (password supersecret). Path: Trader Pools → T1 Renewables Pool → Performance. You should see the same pool name and the latest forecast’s delivery window and generation time.

The demo also contains T1 Industrial Pool. Open it in a second browser tab only if you are comparing portfolios; do not infer a cross-pool total from either book’s KPI cards.

3.3 Read the directory before entering

The table directory and its dated financial verdict are taught in §2. It ranks material financial issues before setup tasks. Empty pools show Not assessed and dashes; an Active status is not a health verdict.

The optional portfolio summary shows the five metrics below. Pool rows always show forecast margin and spill when assessed. A forecast target date describes the model date; the generation timestamp describes when the record was created.

The five portfolio stat cards

Each top card answers one triage question about the whole organisation's trader books. The same labels appear as columns on every pool row — read the card for the portfolio glance, then the matching cell to find which pool is driving it. Full definitions and hour-level proof live in Performance and in the run detail (§4.1, §8).

Demand   |████████████████████| 100 MWh
Delivered|████████████████░░░░|  82 MWh   → Coverage 82%
                     shortfall 18 MWh

Revenue (sell)     +████████████████  R 1.20 M
Costs (buy+whl+…)  −████████████░░░░  R 0.95 M
                   ─────────────────
Margin (fcst)         ███             R 0.25 M

P50 margin |████████████████████| R 1.00 M
P90 margin |██████████████░░░░░░| R 0.72 M
            ←──── MaR 28% ────→   margin destroyed in the adverse case

Supply |████████████████████| 120 MWh
Sold   |██████████████████░░| 111 MWh
                spill 9 MWh → 7.5%

Demand  |████████████████████| 100 MWh
Covered |████████████████░░░░|  82 MWh
                 shortfall 18 MWh → 18%

Other signals on each pool row

Hover any action icon if its meaning is unfamiliar; Strategy is always the scale icon with the "Strategy" tooltip.

Careful. A directory KPI is only as recent as its latest snapshot. If you have just added a contract or changed a rule, open Strategy and use Run Forecast Now before treating the number as a decision. Spill and shortfall can both look "amber-sized" for opposite reasons — read which card (or column) is stressed before you act.

3.4 The shared Strategy shell

Every trader-book route keeps the same context so you do not lose your place while moving from a problem to its evidence:

• ← Trader Pools returns to /org/trader-pools;
• the header repeats the pool name and “Strategy”;
• the tabs are Performance, Allocation, Commercial and Dashboard (#strategy-nav). Allocation stays the lit top tab on Rules, Forecast and Outcomes — those views are not extra room tabs. Allocation opens on Rules. Forecast still lives at …/strategy/book?mode=forecast from the thread strip, a run link, or the old /allocations forward. Each Book view carries the same thread strip, so the wiring, the engine’s plan and the settled month read as one story (§5, §6, §6A).

These are routes within the selected pool, not global worklists:

The old Nominations and Exposure tabs are now the Desk’s Commit step and its exposure section (/exposure forwards; /nominations keeps its URL). The old Pools, Runs, Chains and BESS tabs forward into the Book: /allocations → Forecast, /chains → Outcomes, /bess stays as the battery’s full dispatch page under the Book tab.

Careful. Forecast and Outcomes are the two directions of time in one room. They are not extra tabs on #strategy-nav — Allocation stays the top tab. Forecast (…/strategy/book?mode=forecast) is the strategy engine’s forward ledger for one run. The thread strip at the top of each Book view says which stage you are reading.

3.5 Read badges before acting

Badges here are the small labelled tags on Strategy screens — not the directory KPI values in §3.3. There are two kinds; read both before you treat a number as actionable.

Run and scenario badges

These appear mainly in Allocation’s Forecast run picker (§6) and on a run's detail header (§8). Each run row shows its scenario (what inputs the engine used) and its status (whether the execution finished, failed, or was replaced). They tell you whether a value is a plan, a declaration, a settled result or a safe what-if — read them before comparing two numbers.

Severity badges

These appear on the Performance KPI cards (§4.2) and on the Dashboard (§12). They use the same language everywhere:

If no limit exists, the metric remains ok; green therefore means "within configured limits," not "commercially perfect."

"Read badges before acting" means: check what kind of run you are looking at, and whether a card is warn or breach, before you act on the number.

3.6 A first five-minute tour

1. Open T1 Renewables Pool and confirm the header.
2. Read the Overview cards; if there is no snapshot, select Run Forecast Now.
3. Open Book · Rules and check that supply and demand members look commercially plausible before changing anything; read the outcome badge on each card.
4. Open Book · Forecast; the run picker opens on the newest FORECAST P50 run — this is the evidence behind the Overview.
5. Return to Overview, then use Dashboard only if amber or red needs an hour-by-hour explanation.

Follow along. Seat: trader_1@webill.net (password supersecret). Path: T1 Renewables Pool → Strategy → Runs. You should see a newest-first index where a current completed forecast is not labelled SUPERSEDED.

The next chapter uses the Overview as your daily opening move. When a card points to structural work, go to §5; when it points to a commercial opportunity, go to §9.

4. Overview — run the book

Route: /org/trader-pools/:id/strategy · Tab: Overview

You buy energy from many sellers and sell it to many buyers. Every morning you need one answer for this trader pool: is the book earning, where does the energy go, what can still go wrong, and what should you trade next? Overview is that page. It does not sign a deal. It reads the latest risk snapshot — a sealed forecast result — and turns it into cards you can scan in about a minute.

Think of a merchant’s shop window. The window does not restock the shelves. It shows today’s takings, what is still on the shelf, and the one job to do first. That window is Performance. The storeroom where you wire shelves and baskets is Allocation. The counter where you try a buy or a sell is Commercial. The lifetime accounts are Dashboard.

Where this is taught. How the directory lists every pool: §2. How you create or edit a pool’s licence file: §2B. How you open Strategy for one pool: §3. How you wire members and rules: §5.

How to read the whole page, then what to do

Read top to bottom. Do not start in the middle.

1. Confirm the chip in the top bar names the pool you meant ({pool name} · Strategy). Overview is one book. A healthy neighbour says nothing about this one.
2. Read the first card — the big sentence. That is the period result in one line: the book earns or loses, and often why (spill that never sold, or demand left unmet).
3. Glance at the three tiles under that sentence (Margin@Risk, VaR 95, spill / resale). Then the four Structural position bars beside it. Colour first: rose is a problem, amber is a warning, brand green is inside the tick.
4. Check Concentration. One supplier or one buyer taking most of the book is a different risk from a bad hour.
5. Read Where the energy goes left to right: supply pools → demand pools. Thick green ribbons are good trade. Thick rose ribbons lose money. Amber leaving to Spill · bought, unsold is energy you paid for with no buyer. Rose arriving from Shortfall · unmet demand is load you sold and did not cover.
6. Trust rank 1 on Next best actions, then verify it on the screen the link opens. The list is a first move, not a signed order.
7. Open Long or short, hour by hour only when you need the clock: which SAST hours are matched (green), long / spill (amber), or short / shortfall (rose). Switch Typical day and Tomorrow (P50 supply) to see whether tomorrow’s weather repeats the shape.
8. Use Pool health to name the shelf or basket. Then Where the margin is made and lost for the month × hour, and Forecast said, book settled for whether the plan has been honest.

Good on this page: a headline that earns, spill and shortfall inside their ticks, green ribbons, an empty or quiet action list, and Pool health chips that read ok. Bad: a headline that loses because N% of what it buys never sells, rose gauges, a fat amber spill outlet, rank 1 telling you to sell the spill or cover the shortfall, and breach chips on Solar or Industrial.

Where you go next — only exits that exist on this screen:

There is no Nominations tab in Performance. Nominations keeps its own URL. Reach it from Open Nominations workbench at the foot of Next best actions (#overview-open-nominations), from a ranked Re-optimise → Nominations card when that leak fires, or from Desk · Commit. The workbench’s ‹ Back to Overview (#nominations-back) returns here when you arrived with ?from=overview.

Follow along. Seat: trader_1@webill.net (password supersecret). Path: Trader Pools → T1 Renewables Pool → Overview (or the scale-icon Strategy chip). You should see the subtitle Run the book — position, flow, risk, and the next trade., the intro What the book earns, where its energy goes, what could go wrong, and what to trade next., Explanations beside Run Forecast Now, then the nine cards below.

Data. full_demo · seed 42 · T1 Renewables Pool · last checked 2026-09-18 against StrategyOverviewLive. A seeded book that is long with almost no offtake headlines The book loses R 12.2 m this period because 96% of what it buys never sells. Snapshot Rand on a given machine moves with the latest forecast; the labels and rules below are the product.

The page title is Strategy · Overview. While a forecast runs, a full-screen overlay says Running forecast… and Stay on this page until the run finishes. Tabs and the back link are not clickable. Success flash: Forecast updated. Failure: Forecast run failed. or Forecast run crashed.

If there is no snapshot yet, the nine cards are hidden. You see a dashed empty panel:

No snapshot yet — run a forecast to populate coverage, margin and risk.

Under that empty panel the page still lists Supply and Demand pools by name and member count (N member(s)), or No supply pools. / No demand pools. Those columns are a roster, not health. Press Run Forecast Now (#overview-run-forecast). The button tooltip: Rerun full forecast: base P50 run plus P90 adverse stress run.

Words for this screen (no foreknowledge)

4.0 The book as a marketplace — pools in plain language

Before the cards, you need the two sides Overview assumes. A pool is a named group of members on one side only.

You do not put factories and solar farms in the same pool. You group solar with solar when they share a job (cheap daytime energy), and factories with factories when they share a service (industrial load that should be covered first).

Where this is taught. Creating pools, priorities, limits and members is §5. This section is only the picture Overview’s flow map and Pool health table assume. The hour-by-hour matching diagrams sit with that chart (§4.4).

Four commercial moves. The engine does not invent deals. It only allocates energy inside the deals you already seated.

Layman’s rule. Fill demand in your head first, then buy supply that matches the shape — unless you are deliberately long renewables and hunting offtake. IPP buys own shape and green attributes. Trader buys fill gaps. Offsell to offtakers is the core placement. Offsell to traders is often firm — protect it with Firm supply.

4.1 The position card — what the book earns

On screen: the left card in the first row (#overview-position). There is no folded All five KPIs row and no #overview-kpi-cards. Coverage, spill and shortfall now live on Structural position (§4.2). Margin, Margin@Risk and the spill money live here.

This is the shop-window price. One sentence, a fan if a Monte-Carlo risk run exists, then three tiles.

The headline

The heading (#overview-headline) is chosen in this order from the latest snapshot (MARGIN_R, SPILL_PCT, SHORTFALL_PCT):

Rand is compact: R 412 k, R 12.2 m, −R 12.2 m. On T1 Renewables Pool after the seed forecast, expect the loses … because 96% … never sells line.

Good is earns with no “but spills” / “but leaves unmet”. Bad is loses because — you paid for energy that found no buyer.

The lede and the fan

The grey line under the headline:

• with a fan: P50 margin with the P05–P95 fan from the risk distribution; the dashed tick is break-even.
• without: P50 margin from the latest forecast run; run a risk distribution to see the fan.

The fan (#overview-margin-fan) is a horizontal bar:

Green stroke when P50 ≥ 0; rose when P50 < 0. No Monte-Carlo distribution → #overview-no-fan: No margin distribution yet — the fan appears after a Monte-Carlo risk run. The three tiles still show.

The three tiles

Margin@Risk = (P50 margin − P90 margin) ÷ P50 margin × 100. Example: P50 = R 412,000; P90 = R 298,000 → 27.7%. A fat expected margin that melts in the P90 case is a risk problem, not a success.

4.2 Structural position

On screen: the middle card in the first row (#overview-gauges). Eyebrow: Structural position.

Four bars, always in this order. There are no gauges for firm backing, price exposure, or escalation on this card — those numbers can still raise a next-best action (§4.5).

Each row is: label · bar · value. The bar fills left to right. A black tick marks the limit (labelled limit N%, target N%, or the default words above). Missing value → — and a grey bar (tone: none).

Colours on the bar

Green means the metric is inside your warning threshold — not that the commercial outcome is automatically good. An amber Spill at 7.5% and an amber Shortfall at 7.5% are opposite jobs: one asks you to place midday supply, the other asks you to add supply or reshape sales.

Note. The Trader Pools directory paints rose at fixed triage cuts (§3.3). Inside the book, these bars use your risk limits (§12.6). Trust Overview for operating decisions on the open book.

Take-or-pay used (ToP utilisation): you agreed to pay for a minimum volume even if you do not take it. Low utilisation means you are paying for energy you then spill. The default “good” side is at or above 80%.

Coverage = delivered ÷ demanded. Spill = unsold supply ÷ total supply. Shortfall = unmet demand ÷ total demand. Coverage and shortfall are two views of the same gap: 82% coverage is 18% shortfall on the same numbers.

4.3 Concentration

On screen: the right card in the first row (#overview-concentration). Eyebrow: Concentration.

How much of the book sits with one supplier or one buyer. A healthy margin with one counterparty is still a single-point risk.

Supply HHI and Demand HHI are two meters. The number is the HHI from the snapshot (SUPPLY_HHI, DEMAND_HHI), shown to two decimals, or —. The bar behind the needle is green → amber → rose; the needle sits at HHI × 100% (1.00 is the far right). The note on the right names the largest pool share from this run (Solar 72%) or the hint 1.0 = one supplier / 1.0 = one buyer.

Counterparty exposure lists up to three counterparties from counterparty_breakdown, largest first, each with compact Rand and a share bar (first bar violet, the rest slate). Empty copy: No counterparty exposure in the latest snapshot.

Good is a mid-to-low HHI and a list that is not one name. Bad is 1.00 — one shelf or one basket is the whole book.

4.4 Where the energy goes

On screen: the wide card in the second row (#overview-flow). Title: Where the energy goes.

This is a flow map (a Sankey): shelves on the left, baskets on the right, ribbons between them. The intro on the card:

Supply pools on the left, demand pools on the right, ribbons sized by MWh allocated in the snapshot run and coloured by margin. Energy that finds no buyer leaves as spill; unmet demand leaves as shortfall.

Read it in this order

1. Find left and right. Left nodes are supply pools (Solar, Wind, Flexible, Firm — coloured by type). Right nodes are demand pools (Industrial, Commercial, Municipal, Trader Sales — slate). If two pools share a name, the map adds (supply) or (demand).

2. Read ribbon thickness. Thickness is MWh allocated in the snapshot run (kWh sent ÷ 1,000). A fat ribbon is a big path; a hairline is a small one.

3. Read ribbon colour against the legend on the card:

   Legend swatch	Meaning
   Positive-margin flow	Green — sold above what it cost
   Negative-margin flow	Rose — bought dearer than it sold
   Spill (bought, unsold)	Amber — paid-for energy with no buyer
   Shortfall	Rose outline — unmet demand

4. Hover a ribbon (when the chart has loaded). The label is compact Rand and margin in c/kWh, or bought, never sold / unmet.

5. Find the two outlets. Spill · bought, unsold sits as an extra node: supply ribbons that leave the book unused. Shortfall · unmet demand feeds into demand pools that were not fully served.

6. Read the footer totals (only when there are ribbons): Supply {energy} · Demand {energy}; if spill > 0, bold amber Spill {energy} ({N%}) · {Rand}; if shortfall > 0, bold rose Shortfall {energy} ({N%}).

Empty copy (#overview-flow-empty): No allocation lines in the latest run yet — the flow map fills in after a forecast with seated members.

Good is mostly green ribbons, thin or missing spill and shortfall outlets. Bad is a fat amber river into Spill — that is the same story as a loses because N% never sells headline.

4.4.1 How the sides interact each hour

For each forecast hour the engine roughly does this:

1. Gather available energy in ACTIVE supply members (after caps, nominations, forecasts).
2. Gather requirements in ACTIVE demand members.
3. Serve firm demand from firm-capable supply first.
4. Walk remaining open demand in pool priority, drawing from the supply merit stack.
5. Leftover supply is spill; leftover demand is shortfall.

The flow map is those lines summed for the whole snapshot window. Hour-level proof lives in run detail (§8). Priorities and defaults are §5.1.

4.4.2 Worked example A — midday: sell renewables into offtake

Provenance. Illustrative hour, same conventions as full_demo trader books (c/kWh, MWh). Network losses ignored so the matching arithmetic stays visible.

Situation. 12:00 SAST. Solar has IPP energy. Industrial has factory load. Flexible holds a small trader-in block.

Matching (open demand, no firm this hour)

1. Industrial takes 6 MWh from IPP-A and 2 MWh from IPP-B.
2. Commercial takes IPP-B’s remaining 2 MWh and 1 MWh from the trader-in block.
3. Flexible still holds 1 MWh → spill.

Conservation. Supply offered 6 + 4 + 2 = 12 MWh. Demand served 8 + 3 = 11 MWh. Spill 12 − 11 = 1 MWh ✓.

On the flow map you would see a thick Solar → Industrial ribbon, a thinner Solar → Commercial ribbon, a thin Flexible → Commercial ribbon, and an amber Flexible → Spill · bought, unsold outlet.

4.4.3 Worked example B — evening: firm offtake and trader offsell

Provenance. Same illustrative book, 19:00 SAST.

Situation. Solar is dark. You still owe a firm evening sale to another trader. Industrial is still running.

Matching

1. Firm first: Trader Sales takes 3 MWh from Firm. Remainder 4 − 3 = 1 MWh.
2. Open: Industrial takes 3 MWh Wind + 1 MWh Firm remainder = 4 MWh served.
3. Industrial still needs 1 MWh → shortfall.

Conservation. Supply 4 + 3 = 7 MWh. Demand need 3 + 5 = 8 MWh. Served 3 + 4 = 7 MWh; shortfall 1; spill 0 ✓.

On the flow map: Firm → Trader Sales, Wind → Industrial, a thin Firm → Industrial remainder, and a rose Shortfall · unmet demand → Industrial inlet. Buying more solar would not fix 19:00.

4.5 Next best actions · ranked by rand at stake

On screen: the card beside the flow map (#overview-actions). Title: Next best actions · ranked by rand at stake. A pill shows {N} open.

The panel converts common symptoms into a first move, sorted by Rand at stake (highest first). Rank 1 has an amber surround; the rest are white. Each card has a rank disc, a title, a why, an impact pill, and a deep link.

Empty copy (#overview-actions-empty): Nothing pressing — coverage, spill and assignment all look healthy.

The footer links are always there: Open Simulation Workbench → /strategy/simulations (the Commercial tab) and Open Nominations workbench (#overview-open-nominations) → /strategy/nominations?from=overview. The workbench then offers ‹ Back to Overview.

Impact pills show compact Rand when the engine can price the stake (≈ R 12.2 m), otherwise a plain tag (hygiene, deadline, warranty, structural breach).

An action labelled “add daytime offtake” is a shape hypothesis. Open the Desk, inspect the fit, and test terms before you create a proposal (§9).

4.6 Long or short, hour by hour

On screen: the full-width card (#overview-clock). Title: Long or short, hour by hour.

This is not a separate “Next Best Actions Long or Short” card. Long and short are the clock. The actions card is §4.5.

The intro on the card:

Net position on a typical day of the run: matched energy in green, supply above demand is long (spill, amber), demand above supply is short (shortfall, rose). The solid line is demand; the dashed line is tomorrow’s P50 supply from the suppliers’ weather forecasts.

The period toggle

#overview-clock-period is two buttons, not a URL:

Tomorrow is disabled, at half opacity, with tooltip No supplier weather forecast for tomorrow yet until that extra load arrives. While it loads you may see loading supplier forecast….

How to read Typical day

1. Daylight / Night wash. The legend (#overview-clock-day-night-legend) marks Daylight (06:30–17:30) (sun, amber wash) and Night (moon, grey wash). Hours are SAST.
2. Three stacks, bottom to top in the legend: Matched (green #059669), Long (spill) (amber #f59e0b), Short (shortfall) (rose #e11d48).
3. Solid slate line = Demand (allocated delivered + shortfall that hour, typical-day average).
4. Dashed slate line = P50 supply (tomorrow) when weather exists — overlaid on Typical day so you can compare today’s shape with tomorrow’s expected supply. It does not replace the stacks until you press Tomorrow.
5. Plant now (#trader-plant-chips) sits under that legend, above the chart. Each chip names the IPP site and the last sample time. Press it and Your plant live opens as a scrollable modal on this Overview (#overview-plant-live-modal) — it does not send you to /org/generator-sites/…/strategy/plant. Use it to check the live reading against the P50 tomorrow line. Absent when no seated supplier has a live feed; that is normal.
6. Clip strip (#overview-clip-strip) under the chart: 24 cells, 0h → 23h. Caption: Clip strip · amber = long, the bid desk would clamp · rose = short. Hover a cell: {hh} · long {energy} · short {energy} per typical day. Darker amber means a fatter long hour (the top third of the day’s long peak). Slate means matched with no leftover.

Good on Typical day: a green matched band that follows the demand line, thin amber only at solar noon if you are slightly long, little or no rose. Bad: a tall amber plateau (you bought a shape nobody takes) or a rose evening block (you sold evenings you cannot cover).

How to read Tomorrow (P50 supply)

Press Tomorrow (P50 supply). The stacks are rebuilt as:

• Matched = min(tomorrow P50 supply, today’s typical demand)
• Long = supply above that demand
• Short = demand above that supply

The demand line stays. The dashed P50 line is not drawn again (it is the stack). Use this to ask: if tomorrow’s weather is the P50 we already have, do I still spill at noon or go short at 19:00?

If the button stays disabled, you have no supplier weather yet — stay on Typical day.

4.7 Pool health

On screen: the wide card in the last row (#overview-pool-health). Title: Pool health. Intro: Each pool against its own targets: coverage band, spill and shortfall caps, and its share of the margin.

This is a table, not the old locator “pool cards”. Two section rows — SUPPLY then DEMAND — then one data row per pool (#overview-pool-{id}). Empty section: No supply pools. / No demand pools.

Every table header

How to read a supply row

MWh is allocated sent + spill for that pool. Coverage vs band is blank — supply has no coverage target. Spill / short is the spill share of that pool’s own energy. Chips:

High Solar spill usually means missing daytime demand, not “too much solar forever”.

How to read a demand row

MWh is delivered + unmet. Coverage vs band compares snapshot coverage to the pool’s min–max band (or target ± 10 points, or 90–100). The green wash is the band; the fill is the value (capped at 100% on the bar). Spill / short is the unmet share.

Industrial below its band and Trader Sales at 100% are different jobs. Firm shortfall is rarely acceptable.

The row is a locator, not an editor. Change members and rules on Book · Rules (§5); return here after a new forecast.

4.8 Where the margin is made and lost

On screen: the upper right card in the last row (#overview-heatmap). Title: Where the margin is made and lost. Intro: Month × hour margin from the run’s allocation lines. Green earns, red loses.

A heatmap (#overview-margin-heatmap): columns are months in the run (labels like Oct. '25), rows are SAST clock hours (0h … 23h). Each cell is the sum of margin_r for that month-hour. The colour scale is centred on zero: green earns, red loses, the ends are equal in size so a big win and a big loss are comparable.

Empty copy: No priced lines in the latest run yet.

How to read it. Find the red cells first — those hours lose money. A red evening column every month is a shape problem (firm sales after solar). A red summer-noon column with a green winter-noon is a seasonal spill story. Then open Book · Forecast or Allocations when you need the lines behind a cell. This card does not edit prices.

4.9 Forecast said, book settled

On screen: under the heatmap (#overview-drift). Title: Forecast said, book settled. Intro on the card:

P50 margin at forecast time against the settled trader book, last 6 months. The gap is forecast error you are paying for.

A weather forecast says it will rain. The wet pavement — or the dry one — is what actually happened. This card is that pair, for money.

Forecast P50 is the plan: what the latest P50 (expected-case) risk snapshot said this book would earn that month. Settled book is the receipt: what settlement actually wrote on the trader statement — or, if no buy and sell legs have settled yet, that month’s ACTUAL snapshot.

The chart (#overview-drift-chart) plots both on one pair of axes for the last six calendar months, including this month.

This card does not sign a deal. It does not re-run the forecast. It only asks: has the plan been honest?

Follow along. Seat: trader_1@webill.net (password supersecret). Path: Trader Pools → T1 Renewables Pool → Overview, then scroll under Where the margin is made and lost. You should see #overview-drift, then either Loading settled months…, the two-line chart, or the empty copy below.

Data. full_demo · seed 42 · T1 Renewables Pool · last checked 2026-09-18 against StrategyOverviewLive. Snapshot Rand moves with the latest forecast. The two teaching charts below use the same month labels and compact Rand the card prints. A seeded book that is long may show a large gap, or No month has both a forecast and a settled statement yet. The labels and rules are the product.

Where this is taught. The month’s buy and sell legs behind Settled book: §6A. How a snapshot is sealed: §4.10. The IPP card Forecast said, settlement did is a different screen — last month’s buyer carve, not this book’s margin (V4 §8.1.8).

Words on this card

The gap is error, not a bill on its own. Settled below forecast means you planned on money you did not get. Settled above forecast means you earned more than the plan — still a miss, but a happy one.

How to read the chart

1. Find the legend under the plot. Two names: Forecast P50 and Settled book.

2. Read the colours (Overview uses the shared line palette). Forecast P50 is first — brand green #059669. Settled book is second — sky #0ea5e9.

3. Read left to right. Each tick is one calendar month. Hover a month to see both Rand figures.

4. Ignore one line alone. A falling settled line is only a problem relative to the forecast. Two lines that fall together mean the plan saw the drop.

5. Look at the gap. Do the lines travel together, or does one wander off? Does the space grow as you walk toward this month?

6. Read the verdict under the chart, once extras have loaded:

   Verdict	When
   Settled margin runs {Rand} below forecast on average over {N} month(s).	Average gap is negative — the plan was optimistic
   Settled margin runs {Rand} above forecast on average over {N} month(s).	Average gap is positive — the plan was pessimistic
   No month has both a forecast and a settled statement yet.	Lines may exist, but no month has a pair, so there is no average

A missing dot in one series is normal. That month has a forecast or a settlement, not both. The verdict ignores it.

Loading (page just connected): Loading settled months… — the chart is a slow extra. Wait. Empty (#overview-no-drift): No forecast snapshots or settled statements in the last 6 months yet. Press Run Forecast Now, or wait until a month has settled.

A good chart — the forecast was honest

A good chart is not “the book earned a lot.” It is “the plan and the receipt stay close.” Small gaps that flip sign (a little above, a little below) are weather noise. The gap does not grow.

Why the good one is good. The weather report was close to the pavement. A May beat and an April miss cancel. You can use the headline on §4.1 as a planning number.

Why the bad one is bad. The report kept promising rain. The pavement stayed dry — then the book crossed R 0 and kept losing. A common cause on a long trader book: the forecast still prices offtake you have not seated, so P50 margin stays fat while the ledger records spill. T1 Renewables Pool’s loses because N% never sells headline is that story in one sentence. This chart is the six-month proof.

What to do next

Book · Outcomes draws a cousin of this chart for the same six months. Dashboard keeps the long view (§12).

4.10 What happens when you run (or re-run) the forecast

Run Forecast Now (#overview-run-forecast) is the on-demand twin of the nightly strategy cycle for this book. It asks the engine:

Given everything seated on this book today — members, prices, caps, rules, forecast shapes, network losses, and the battery if you have one — what does the next year look like in the expected case, and how much of that margin melts away if renewables under-deliver?

The answer is frozen as a risk snapshot. Overview, Dashboard, and the directory chips all read that snapshot. No snapshot → the empty panel in the chapter open. A successful press replaces today’s snapshot for this book.

Same cycle, two triggers. Nightly (about 02:30 SAST, every active book) and Run Forecast Now (this book, now) both call the Margin-at-Risk pass: P50 base + P90 adverse + snapshot. The nightly rhythm is V3 §18.3.

The Book · Rules amber banner Forecast is behind this board is taught in §5.6. A simulation in Commercial is not a forecast re-run (§9).

4.10.1 When you should press it — and when you should not

4.10.2 The pass in seven steps

No conveyor-belt diagram — the order is the list.

1. Window. Today → a rolling 12-month half-open window. Example: press on 23 Jul 2026 → [2026-07-23, 2027-07-23). Each hour is its own match (§4.4.1).
2. Inputs. Freeze active pools and members, prices and caps, forecast grade, network edges, battery, period. Fingerprint: inputs hash. Same hash → same lines.
3. P50 pass. Allocate every hour (firm first, then open priority; tag SPILL / SHORTFALL; optional battery pass). Store a FORECAST / P50 run. This becomes the current forecast position.
4. P90 pass. Same period; every renewable member swaps to P90. Tagged SIMULATION / P90 so it never becomes “the” current forecast.
5. Margin-at-Risk. MaR (R) = P50 margin − P90 margin; MaR (%) = MaR (R) ÷ P50 margin × 100. That percentage is the Margin@Risk · P90 adverse tile.
6. Snapshot. One dated row: portfolio metrics, per-pool breakdown, breaches against active limits.
7. Screens. Overview reloads the nine cards. Dashboard reads the same snapshot. The previous equivalent FORECAST run is marked superseded only after this press completes.

While the overlay is up, keep treating the previous snapshot as the decision record. Only Forecast updated. means steps 1–7 finished.

4.10.3 P50 and P90 in plain language

Renewable forecasts are graded by exceedance:

Common mix-up. “P90” is not “ninety percent — a lot.” It is the low case you are still confident of clearing. P90 ≤ P50 ≤ P10 for generation.

Toy plant: declared P50 = 43,800 MWh/yr, solar-like P90/P50 = 0.94 → P90 ≈ 41,172 MWh/yr. On a book, the adverse pass shrinks every renewable hourly shape, then re-matches the whole marketplace. Firm supply is not shrunk the same way.

Snapshot the Overview actually shows

Illustrative trader book on 1 Aug 2026 (rounded whole Rand):

MARGIN_R              = 412 000        ← from P50 run
MARGIN_AT_RISK_PCT    = 27.7           ← (412k − 298k) ÷ 412k × 100
COVERAGE_PCT          = 82.0
SPILL_PCT             = 9.3
SHORTFALL_PCT         = 1.8
pool_breakdown[Industrial].COVERAGE_PCT = 75.0
breaches = [ MARGIN_AT_RISK_PCT warn if limit warn=25 ]

On T1 Renewables Pool after the seed, expect the opposite commercial shape: a loses R 12.2 m headline and spill near 96% — same cards, different numbers.

4.10.4 Re-run vs first run

Press twice with no book changes: new run ids, same inputs hash and KPI values. Press after seating a priced PPA: hash, Margin and often MaR move.

4.10.5 Operator checklist

1. Finish the structural change on Book · Rules, the battery or limits.
2. Press Run Forecast Now here or on the Pools banner.
3. Stay on the page while the overlay is up.
4. When Forecast updated. appears, read the headline and rose gauges first, then rank 1.

Follow the lines. Allocation mechanics: §7. Runs list: §6. Dashboard: §12.

4.11 Explanations and other chrome

Explanations (#overview-explanations-open) opens a right-hand drawer (#overview-explanations-drawer). Eyebrow: Trader Overview. Title: Explanations. Above the terms sits How to Read the View (#overview-how-to-read) — collapsed until you press #overview-how-to-read-toggle. It is the short in-product reading order: position, structural position, concentration, energy flow, actions, hour-by-hour, pool health, margin heatmap, forecast versus settled. Intro: Abbreviations and terms used on the trader-pool Overview, in alphabetic order. Close with the X (#overview-explanations-drawer-close), the backdrop, or Escape. The list is A–Z from BESS (Battery Energy Storage System) through VaR 95 (Value at Risk). It does not include SAWEM — that word is not on this Overview (it lives on the Bid Desk).

Plant now chips (#trader-plant-chips) sit under the Long or short legend, above the chart — not above the intro. They appear only when a seated supplier has a live telemetry feed. Each chip shows the IPP site name and last sample time. Press the chip: Your plant live opens as a scrollable modal (#overview-plant-live-modal, branded header, site name as the title). Close with the X (#overview-plant-live-close), the backdrop, or Escape. You stay in Performance. Absent on a book with no live feeds — that is normal, not an error.

Busy overlay (#strategy-busy-overlay): Running forecast… / Stay on this page until the run finishes.

This is the morning pattern: Overview detects, Book / Desk / Dashboard change, a fresh snapshot confirms. Keep the two-sided picture from §4.0 when Pool health points at Solar spill versus Industrial shortfall — those are opposite shelves.

Where this is taught next. Book has its own Explanations drawer with a different How to Read for Rules, Forecast and Outcomes (§5.8, §6.8, §6A.4).

5. Book · Rules — how the book is wired

Route: …/strategy/book opens here (…/strategy/pools) · Tab: Book (#strategy-tab-book stays active) · Mode: Rules (#book-stage-rules on the thread strip; Outcomes and Forecast are the other stages, not extra #strategy-nav tabs.)

What it is for. Pools are how you tell the engine what your book is. A pool is a named group of members on one side — supply (sources) or demand (sinks) — with a stated purpose, a priority, and side-specific limits. The engine only sees ACTIVE pools; a PPA that isn't a member of any pool doesn't exist as far as allocation is concerned.

Where the marketplace picture is taught. Why supply and demand pools exist, when you buy from an IPP vs another trader, and when you offsell to an offtaker vs a trader — with midday and evening worked examples — is §4.0; the hour diagrams sit with the Overview flow map (§4.4). This chapter is the build and edit screen: create pools, seat members, set priorities and rules.

Why pools at all, instead of just a list of contracts? Because allocation decisions are made between groups: solar as a class serves daytime load first; firm supply as a class is reserved for firm obligations; spill from renewables is routed to the flexible pool for tactical use. Pools carry those group-level policies (priorities, spill destinations, coverage targets, rules) so you never have to encode strategy contract by contract.

The Book’s thread, and what this mode adds to the board

Allocation on the strategy row opens Rules. The top underline (#strategy-nav) stays Operations · Performance · Allocation · Commercial · Dashboard — Allocation stays lit. Rules, Forecast and Outcomes switch on the thread strip under that bar, not as extra room tabs.

Every Book view still opens with the same thread strip: three stages — Rules · how the book is wired → Forecast · what the engine will do → Outcomes · what actually settled — each with a verdict pill (on the demo: 4 caps breached → September P50 · −R 12.2 m · spill 95.8 % → August · −R 381 k · not balanced). Clicking a stage opens that view. Under it the issues rail ranks everything wrong in the book — caps breached, a month that does not balance, a nominated run still running, a failed run, coverage outside its band, draft pools with no members — and each item points at the mode that fixes it.

Then the mode’s headline, written from the numbers: “The cascade sells into 4 demand pools in priority order — and supply outweighs what they can take 24 to 1. Whatever Solar, Wind, Flexible and Firm bring past 4 % will spill.”

The board itself is two columns with a wiring gutter between them. SUPPLY pools sit left in merit order, DEMAND pools right in priority order. Three things are new on the cards:

• an outcome badge at the foot of each card — the pool’s cap or band as a bar with the limit tick, filled with what the latest forecast actually did to it: green inside the cap or band, amber outside the band, red when a cap is broken. The wiring problem appears on the card you edit;
• wires in the gutter (#book-wires) — coloured joiners from a supply card to a demand card. The colour is the supply shelf (Solar amber, Wind light blue, and so on). The word on the curve is the rule (shape match, price ≥ 12.5 c, …). A dashed red curve is leftover supply the latest forecast could not sell. The full reading guide is §5.7.

There is no three-card “next moves” strip under the board. Reorder supply on the left column; coverage bands live on the demand-card badges; spill and shortfall are on the issues rail. When the latest forecast leaves more than 20 % of supply unsold, a single line (#book-offtake-cta) offers Find offtake → Desk — the same Desk door Overview already ranks under Next best actions.

Everything else on this board — creating pools, seating members, rules, book rates, the unassigned-contracts strip and the amber Forecast is behind this board banner — is unchanged and taught below.

Above the board:

• Book rates — book-wide default spill value and shortfall top-up (c/kWh). Member assign prefills from these when set; blank member fields inherit them at run time (§5.3).
• Forecast is behind this board (amber banner) — appears when members, rules, priorities / caps, book rates, or an ACTIVE → ARCHIVED pool change the book after the last sealed forecast. [Run Forecast Now] on the banner runs the same cycle as Overview (§4.10, §5.6); the banner clears when that seal succeeds.

5.1 Creating pools — and what the defaults mean

• [Seed default pools] (only shown while you have none) creates the standard Phase-1 set of eight pools. These are not arbitrary — their priorities and limits encode a sensible starting strategy:

  Supply side (priority = order the allocator prefers them, lower first):

  Pool	Priority	Max spill	Purpose
  ☀ Solar	10	15%	Cheap daytime energy; spill-tolerant.
  ☁ Wind	20	12%	Night coverage and shape diversity.
  🎚 Flexible	30	20%	Nomination-based and tactical; receives the spill from Solar and Wind.
  🔒 Firm	40	5%	Dispatchable/firm blocks reserved for firm obligations.

  Cheapest-and-least-firm first, most-expensive-and-most-precious last: the allocator drains solar before it touches your firm block, and the firm block is held back for the customers who contractually must be served.

  Demand side (priority = order in which demand pools get access to supply):

  Pool	Priority	Coverage target (min–max)	Max shortfall	Firm
  🔧 Industrial	10	75% (70–90)	5%	—
  🏬 Commercial	20	55% (50–60)	10%	—
  🏛 Municipal	30	65% (60–70)	10%	—
  💵 Trader Sales	40	100% (100–100)	0%	✓

  Note the deliberate asymmetry: Trader Sales has the last priority number but is firm — and firm demand is served first, out of turn, from FIRM supply (§7.1). The priority ordering only governs the open (non-firm) demand queue.

  How supply priority is applied. For every hour, the engine builds a supply merit stack from energy that is actually available in that hour. Lower numbers go first. The ordering is:

  1. supply-pool priority;
  2. buy price, cheapest first, for members in the same pool;
  3. member priority within that pool; and
  4. member ID as a final deterministic tie-break.

  FIRM-pool supply is treated specially: it is offered to firm demand before the open merit stack is walked. Any FIRM supply left after firm obligations have been met becomes the last source available to open demand. Capacity caps, nominations and the hour's forecast are applied before the stack is built, so priority never creates energy that is not available.

  How demand is matched and prioritised. The engine first separates firm demand from open demand. It serves all firm demand first, ordered by demand-pool priority and then member priority. It then serves open demand in the same lower-number-first order. Each demand tier draws from the eligible supply stack until its requirement is met or supply is exhausted. Demand members with the same pool and member priority share the available supply pro rata, using their allocation weights, rather than the first member taking everything. Coverage targets and maximum-shortfall settings are monitoring and risk thresholds; they do not override this hourly service order.

  Worked example — one hour, ignoring network losses for clarity.

  Available supply is Solar A 6 MWh at 65 c/kWh in the priority-10 Solar pool, Solar B 4 MWh at 70 c/kWh in the same pool, Wind 5 MWh at priority 20, and Firm 4 MWh at priority 40. Demand is Trader Sales 3 MWh firm at priority 40, Industrial 8 MWh at priority 10, Commercial 5 MWh at priority 20, and Municipal 5 MWh at priority 30.

  1. Trader Sales is firm, so it takes 3 MWh from the Firm pool first, leaving 1 MWh of Firm supply.
  2. Industrial is the first open-demand pool. It takes Solar A's 6 MWh and 2 MWh from Solar B.
  3. Commercial is next. It takes Solar B's remaining 2 MWh and 3 MWh from Wind.
  4. Municipal is last. It takes Wind's remaining 2 MWh and the final 1 MWh of Firm supply. It is left with a 2 MWh shortfall.

The result is auditable: firm demand was protected first, open demand followed its priority queue, and each demand drew from the highest-priority (lowest-numbered) eligible supply before more precious Firm energy was released. Within Solar, the cheaper Solar A energy cleared before Solar B.

• [+ New pool] opens a viewport-fitting modal (sticky header and footer; the middle scrolls). Fill every required field, then Save pool. The tables below list every field on that form: what it means, a concrete sample, and why you enter it. Choosing an archetype Type can prefill several limits — §5.1.4.

New pool — fields always on the form

The priority strip under Default priority lists other pools on the same side so you can pick a number that slots correctly (e.g. between Solar 10 and Wind 20).

New pool — supply-only fields

Shown when Side = Supply.

Sample story. Solar produces 100 MWh; 82 MWh are allocated → spill = 18%. With Max spill 15%, the book flags a breach. If Spill destination = Flexible, those 18 MWh appear on Flexible for follow-up while still unsold.

New pool — demand-only fields

Shown when Side = Demand.

Sample story. Industrial demand 100 MWh, delivered 65 MWh → coverage 65% (below min 70%) and shortfall 35% (above max shortfall 5%) — both risk conditions fire. These percentages are measurement and alert settings; they do not change the allocator's hourly priority walk.

Required to activate later. After save the pool is DRAFT. Activation still needs purpose, at least one member, and the side's key limit (Max spill % on supply, Coverage target % on demand) — see the card actions in §5.2.

Type suggestions on New pool

When you create a pool by hand, you do not have to invent the risk numbers. Choosing a known Type on + New pool prefills the same archetype limits that Seed default pools uses (the tables at the top of this section). A green hint under Type summarises what was filled.

Worked example — New Solar pool. Side = Supply, Type = Solar → the form fills Max spill 15%, priority 10, and purpose "Cheap daytime energy; spill-tolerant." The hint reads roughly: Suggested for Solar: max spill 15%, priority 10. Editable — change any value to override.

Worked example — New Industrial pool. Side = Demand, Type = Industrial → Coverage target 75%, band 70–90%, max shortfall 5%, priority 10.

Override rules (so suggestions stay helpful, not sticky)

1. Suggestions run only on New pool, never when you Edit an existing pool (your saved limits stay put).
2. Edit any suggested field and that value is yours — changing Type later will not overwrite a field you already customised.
3. If you leave a field on the previous type's suggestion (or still blank / priority still 100), switching Type replaces it with the new archetype.
4. Custom clears the green hint and does not invent numbers; keep or clear prior values yourself.

Use Seed default pools when you want all eight archetypes at once. Use + New pool + Type suggestions when you only need one more shelf or basket (for example a second Solar-shaped CUSTOM is still blank — pick Solar if you want Solar's spill tolerance).

How to set the supply-side spill fields.

Max spill % is the largest share of a supply pool's energy that may remain unsold before the dashboard flags it. It is a risk tolerance, not a cap on generation: the engine still records every unsold MWh as spill.

Spill destination pool is the pool where that spill is routed for tracking and tactical follow-up. It does not make the energy sold or create revenue; it gives the operator a clear place to see energy that could be nominated, sold, or used to charge a battery.

Example — Solar Pool. A solar pool produces 100 MWh in a period and the engine allocates 82 MWh. Its spill is (100 − 82) ÷ 100 = 18%. If Max spill % is 15%, the 18% result breaches the pool's tolerance. If its spill destination is Flexible Pool, the 18 MWh is shown there for follow-up; it is still unsold energy until an actual commercial or battery action uses it.

How to set the demand-side coverage and shortfall fields.

Coverage target % is the operating aim for the share of demand that the book should deliver. Coverage min % and Coverage max % form the acceptable band around that aim: below the minimum is under-coverage risk; above the maximum signals over-procurement risk. Max shortfall % is a separate limit on unmet demand. These are dashboard and risk thresholds; they do not change the engine's hourly priority walk.

Example — Industrial Pool. Set a coverage target of 80%, an acceptable band of 70–90%, and a maximum shortfall of 30%. For 100 MWh of demand:

• 78 MWh delivered means 78% coverage and 22% shortfall. Coverage is inside the band and shortfall is inside its limit, although the operating target was missed.
• 65 MWh delivered means 65% coverage and 35% shortfall. Coverage is below the minimum and shortfall exceeds its limit, so both risk conditions are flagged.
• 94 MWh delivered means 94% coverage. It is above the maximum, signalling that the pool is being covered more heavily than its commercial policy allows.

Choose the target, band and shortfall limit together so they express one coherent commercial policy for the pool.

5.2 The pool card

Each card shows the pool's type icon, status badge (DRAFT / ACTIVE / ARCHIVED), purpose, side-specific figures (supply: spill ≤ X%; demand: target X% and a 🔒 firm badge), member count and its member list (each with its priority and a remove ×). Every member line starts with the organisation that owns that site or trader block, then the site or pool name, as Organisation · Site (the same pattern as Deal Room counterparties). A contract-only member (saved with physical_ppa_id) still resolves to that counterparty organisation and site or pool. Card actions:

The operational summary also shows the pool's priority. Lower numbers are called first in the allocation stack; 100 is last. The Member compatibility panel shows the accepted member-class chips (for example, Solar, Firm or Industrial) and the match policy configured in the pool form (§5.1): Strict means a member must share at least one accepted class to be offered or added; Approved override permits an intentional exception while retaining the configured classes as the pool's commercial rule. Use the panel to check the admission rule before adding a member (§5.3).

Reorder members from the card. The member list is grouped into priority tiers, with the lowest priority shown first. Within a shared priority tier, members are shown by ascending pro-rata weight; members with a blank weight (which use contracted volume) follow the explicit weights. This makes the allocation order and every pro-rata group visible without opening another screen.

Click Edit order beside a member to change its Priority or Pro-rata weight. Save the change and the card immediately moves the member into the appropriate priority tier and re-sorts that tier by weight. Priority controls which tier is considered first and cannot exceed 100; pro-rata weight controls the split only among members in the same priority tier, must be greater than zero, and cannot exceed 10. Leave the weight blank to use the member's contracted volume instead.

• + member — assign an unassigned executed PPA to this pool (§5.3); greyed out until one exists
• Rules — manage allocation rules (§5.4)
• Edit — reopen the pool modal
• Activate (DRAFT and ARCHIVED pools, #activate-pool-{id}) — the engine only allocates ACTIVE pools. On a DRAFT card you also have Archive — go live, or retire the shelf without ever putting it in a run. Press Activate on a draft to bring it into the book, or on an archived card to put that shelf back into runs. You do not create a second pool of the same type: an archived Solar still occupies Solar on Select a type. Activation is validated against three requirements: a non-empty purpose, at least one member, and the side's limit set (max spill % for supply, coverage target % for demand). If something is missing you get the exact reason as a flash, e.g. "Cannot activate: coverage_target_pct must be set". Success flash: Pool activated. The point of the gate: a pool without a target can't be measured, and a pool that can't be measured shouldn't be in your book. Activating (or any other structural edit that changes the board fingerprint) marks Overview / Dashboard stale until you re-run the forecast (§5.6).
• Archive (DRAFT and ACTIVE pools, #archive-pool-{id}) — retires the pool from future runs. On a draft this is how you put the card away without activating it first. Past runs keep their lines (immutability, §1.3) so history is unaffected. The card stays on the board with an ARCHIVED badge. Press Activate on that card when you want it in the book again. Archiving is a structural change: the amber Forecast is behind this board banner returns until you press Run Forecast Now.

5.3 Adding members

On the trader Pools board, a member is an unassigned executed physical PPA (SIGNED or ACTIVE) that you assign to a pool. Prospect generator sites, offtaker sites and trader blocks are not addable from + member here — conclude the contract first, then assign it.

+ member is enabled only when the trader has at least one unassigned executed PPA (the same set that feeds the unassigned-contracts strip in §5.5). If none exist, the button is greyed out with "No unassigned contracts — conclude a PPA first".

Click + member on the target pool. The modal ranks those unassigned contracts for that pool: side must match, the counterpart must pass the pool's Strict class compatibility, and only matches with a suitability score above 50% are listed. Score chips use amber (fair, >50–75), light green (good, >75–90) and green (strong, >90). Select a row to open the configuration drawer (priority, pro-rata, caps, spill / top-up); save creates a membership with physical_ppa_id only.

If every unassigned contract scores 50% or below for this pool, the list is empty — use the strip's type suggestion, or open + member on a better-matching pool type. The strip's [Accept suggestion] / [Choose pool…] path remains the one-click / pick-pool onboarding for the same contracts.

Per-member fields, and why they exist:

• Label — optional override on the add-member form. The Rules card prefers organisation · site or pool name when those are known (see §5.2); otherwise it falls back to this label or to PPA {agreement ref}.
• Priority — merit order within the pool, lower first. Between two solar PPAs, the cheaper/preferred one should carry the lower number.
• Pro-rata weight — when two members share the same priority, demand is split between them proportionally. If you leave it blank the weight defaults to the member's contracted volume, which is the settlement convention — so same-priority members share load the same way the settlement allocator would.
• MW cap — hard hourly ceiling for this member (also capped by the PPA's contracted capacity, whichever is lower).
• Nominated — marks a supply membership whose volumes come from submitted nominations rather than a forecast shape (typically a flexible / trader-fill contract). The engine then reads nomination rows for each hour, and ticking it enrols the member in the Nominations workbench (§10). The contract-assign + member drawer on trader Pools does not offer this toggle today — nominated flexible members in the demo are seeded; edit membership order / future config covers day-to-day changes.
• Spill value (c/kWh) — supply pools only: what a spilled kWh is still worth to you (a recovery or resale price). The trader book carries a Book rates default (Pools header); assigning a contract prefills from that book default when set, otherwise from the PPA's spill credit / price. Leaving the member field blank inherits the book default at run time (then zero if the book has none) — spilled energy is only a pure loss when both are unset. An explicit member value stays an override when book rates change. This same number is what charging a battery from spill "costs" (§11.2).
• Eskom top-up / Shortfall top-up (c/kWh) — demand pools only: what it costs to cover this member when your book falls short (grid / replacement energy). On the member form the field is labelled Shortfall top-up. Prefill prefers the book's shortfall top-up default (Book rates on Pools); otherwise the contract shortfall penalty / energy price. Blank inherits the book default at run time. Used to price firm shortfalls (§7.2), the nomination optimiser's cost of under-nominating (§10.2), and the battery's value when it backs a firm shortfall (§11.2).

Worked example — priority, then pro-rata. In one hour, the pool has 10 MW of demand to serve. The engine walks priority tiers in ascending order; it does not pro-rate across different priorities.

Member	Priority	Pro-rata weight	MW cap	Allocated MW
Solar PPA A (lowest-cost)	10	—	3.00	3.00 — served first, up to its cap
Solar PPA B	20	1	4.00	1.75
Solar PPA C	20	3	6.00	5.25
Solar PPA D	30	—	2.00	0.00 — no demand remains

PPA A takes 3.00 MW first, leaving 7.00 MW for the priority-20 tier. PPA B and C then split that remainder in the ratio 1 : 3: 7.00 × 1 ÷ (1 + 3) = 1.75 MW and 7.00 × 3 ÷ (1 + 3) = 5.25 MW. Their caps do not bind, so the split stands. PPA D is considered only after the priority-20 tier; by then the 10 MW requirement is met. If B and C had blank weights, the engine would use their contracted volumes instead — for example, 2 MW and 6 MW would produce the same 1:3 split.

Members can be removed from the card's member list (trash control, with confirm). Removal destroys the membership and returns the executed PPA to the Unassigned contracts strip (§5.5) so you can place it on another pool. Historical allocation lines keep their kWh and prices; membership tags on those lines are cleared so the contract is free to reassign. Removing a member (or changing its priority / weight / caps) also raises the forecast-needed banner (§5.6).

5.4 Rules

Per pool, an ordered rule list that steers how the allocator serves it. Rules refine the default merit order — you only need them where the default (priority + price) doesn't express your contract or strategy.

Opening Rules. Click Rules on a pool card. The modal lists existing rules (reorder ▲/▼, enable/disable, delete) and a side-filtered picker: only rule types valid for that pool's side appear, and types already on the pool are omitted until you delete them. Each available type is a card with its name and a short description (the same intent as the table below). Pick a card, fill parameters when needed, then add.

Supply-side picker: PRIORITY, PRO_RATA, MW_CAP, NOMINATION, SHAPE_MATCH.

Demand-side picker: PRIORITY, PRO_RATA, MW_CAP, PRICE_OPTIMISE.

(PRICE_OPTIMISE is demand-only in the picker; NOMINATION and SHAPE_MATCH are supply-only.) A supply Shape match that names a Target pool is drawn as a coloured joiner in the gutter — the click path is §5.7.2; how to read the colours is §5.7.

5.4.1 Worked rule set — three hours, all six rule types

The following compact book shows how the rules combine. Network losses are ignored so the MWh remain easy to follow. Across the relevant pool Rules modals, the active rows are:

5.4.2 Example 1 — priority, pro-rata and a pool cap

At 12:00, Solar has 10 MWh. Municipal normally follows Industrial (default priorities 30 and 10), but its PRIORITY 5 rule moves it to the front for this run. Municipal contains two members: Water Works needs 4 MWh at weight 1; Clinic District needs 8 MWh at weight 3. The Municipal pool's MW_CAP 8 rule limits the pool before the tier is shared.

The run therefore records 8 MWh on the Municipal edge, split 2/6, plus 2 MWh to Industrial. Municipal has 4 MWh of unmet demand and Industrial has 6 MWh unmet. If Water Works needed only 1 MWh, weighted water-fill would give it 1 MWh and hand its unused share to Clinic District, subject to Clinic's own requirement.

5.4.3 Example 2 — shape dedication and margin protection

At 23:00, Wind has 10 MWh available at 75 c/kWh and wheeling is 15 c/kWh. Industrial's night shift needs 6 MWh and pays 110 c/kWh; Commercial needs 5 MWh and pays 85 c/kWh. The off-peak SHAPE_MATCH declares Wind → Industrial as the dedicated pairing, so the first 6 MWh goes to the night shift. The remaining 4 MWh reaches the Commercial edge, where PRICE_OPTIMISE tests the margin:

Commercial margin = 85 − 75 − 15 = −5 c/kWh
−5 < required +5 c/kWh → skip the allocation

Commercial therefore takes 0 MWh from Wind and falls to its grid backstop; the unused 4 MWh Wind remains available for another eligible buyer or becomes spill. If Commercial were firm, the margin floor would not block it: firm service overrides PRICE_OPTIMISE.

Current-phase note. The SHAPE_MATCH row stores and exposes the off-peak dedication as auditable intent. Where strict shape enforcement is not enabled in the deployed engine milestone, ordinary merit order still decides the physical allocation; the other calculations in this example remain the rule's intended enforced outcome.

5.4.4 Example 3 — nomination is a ceiling, not a promise

At 19:00, the Flexible member has a submitted nomination of 7.5 MWh, actual available supply of 9.0 MWh, and matched demand of 8.5 MWh:

allocated = min(9.0 available, 7.5 nominated) = 7.5 MWh

The run allocates 7.5 MWh, records 1.5 MWh spill from availability above the nomination, and records 1.0 MWh shortfall against the 8.5 MWh demand. On a poor evening with only 6.0 MWh available, the same rule allocates 6.0 MWh: nomination limits upside but never manufactures energy.

Rules are applied in precedence order — use the ▲/▼ buttons to reorder, the Enable/Disable toggle to switch one off without deleting it (it greys to half opacity), and 🗑 to delete. New rules append at the bottom. Every run line records which rules produced it (the Rules column in run detail, §8.5), so you can verify a rule is actually biting. Adding, removing, reordering or toggling a rule marks the forecast stale until you re-run (§5.6).

5.5 Unassigned contracts (the onboarding strip)

An amber strip below the board lists every executed PPA of yours that is not in any pool. Each row shows a Supply or Demand side badge (buyer vs seller leg for your trader), plus a type suggestion ("wind IPP, as-generated → Wind pool"):

• [Accept suggestion] — one click creates the membership in the suggested pool. You get "Contract assigned to ." — or, if you haven't created a matching pool yet, "No pool yet — create one first."
• [Choose pool…] — opens the member modal with the contract fixed and ranks same-side pools by suitability %. Score chips: red / disabled at 50% or below; amber >50–75; light green >75–90; green >90. Select an eligible pool, configure priority / caps, then save.

The strip disappears when everything is assigned, and every newly executed contract reappears here until you place it. Removing a member from a pool (§5.3) puts that contract back on the strip. Treat it as a to-do list, not information: an unassigned contract is real money the engine cannot see, so your margin and coverage numbers are wrong until the strip is empty.

5.6 When the Pools board asks you to re-run the forecast

Overview and Dashboard read a sealed risk snapshot. That seal stores a fingerprint of the book's structure (pools, memberships, rules, book rates, pool status). When the live Pools board no longer matches the most recently sealed forecast, an amber banner appears at the top of Pools:

Forecast is behind this board — Members or rules changed since the last run. Overview and Dashboard still show the previous book — run a forecast to refresh them.

[Run Forecast Now] on the banner (and the same control in Performance) starts the Margin-at-Risk pass described in §4.10. While it runs:

• a full-screen spinner overlays the strategy area ("Running forecast…" / "Stay on this page until the run finishes.");
• strategy tabs and the Trader Pools back link are not navigable;
• the button shows a spinning icon and stays disabled.

When the seal succeeds, the flash "Forecast refreshed." appears, the overlay clears, and the amber banner disappears — Overview / Dashboard now match this board.

The banner returns when you:

• add, remove, or re-order / reconfigure a member;
• add, remove, reorder, or toggle a rule;
• change book rates or other fingerprint fields on pools / memberships;
• archive an ACTIVE pool (or otherwise change pool status in a way the fingerprint tracks).

It does not stay up merely because an older or future-dated snapshot row exists without a fingerprint — the board compares against the most recently sealed FORECAST.

5.7 The coloured joiners between Supply and Demand

The two-column board is a pair of warehouse aisles: Supply shelves on the left (sources, in merit order), Demand baskets on the right (buyers, in priority order). Between them sits a narrow strip of coloured joiners — the wiring gutter (#book-wires inside #book-cascade-board). Think of jumper cables between shelves. The colour of the cable tells you which supply shelf it left. The tag on the cable tells you the rule that named that destination. A dashed red cable is not a rule: it is leftover energy the latest forecast could not sell — spill.

Follow along. Seat: trader_1@webill.net (password supersecret). Path: T1 Renewables Pool → Strategy → Allocation → Rules. URL: /org/trader-pools/01a0ab54-e46b-7dde-b45f-73d119213c75/strategy/pools. Widen the window past the phone breakpoint — the gutter is hidden on a stacked (mobile) board (hidden until md). You should see at least one solid coloured curve with a short label, and — once a forecast has spilled — a dashed red rest → spill · N %.

What a solid joiner is. The engine only draws a solid curve when a supply pool has an active allocation rule that names a living demand pool (target_pool_id set, and that demand card is not ARCHIVED). One curve per such rule. The arrow-head points at the demand card. The curves are pictures, not buttons: they are aria-hidden SVG. You do not draw them by dragging between cards. You add a Shape match rule on the supply card and name a Target pool — the click path is §5.7.2.

Colour is the supply type, not the rule. This is the usual misread. On T1 an amber curve labelled shape match looks as if amber means shape-match, and a light-blue curve labelled price ≥ 12.5 c looks as if blue means price. Those colours belong to Solar (#f59e0b amber) and Wind (#0ea5e9 light blue). Put the same shape-match rule on Wind and the curve stays light blue; the word still says shape match.

Demand cards themselves use a near-black slate (#1e293b); they do not paint the joiners. Archived supply shelves are omitted from the gutter, so their rules do not draw.

The word on the curve is the rule. The label is a short gloss of rule_type (and its parameters):

Rules without a destination demand pool stay on the card’s Rules modal only. They do not appear in the gutter. Pool-wide PRIORITY / PRO_RATA / MW_CAP / NOMINATION on a demand card likewise stay on that card.

The dashed red joiner is leftover supply. When a forecast allocation run exists and some of that run’s supply was unsold, the gutter adds a dashed rose (#e11d48) curve along the bottom, labelled rest → spill · N %. N is the unsold share of total supply in that run, whole percent. No forecast, or a forecast with zero spill, means no dashed curve. The dashed line does not point at a demand card: spill has no buyer yet. When that leftover is more than 20 % of supply, Find offtake → Desk (#book-offtake-cta) sits under the board — not a trio of coaching cards.

Empty and hidden states.

• No solid joiners. No supply rule names a non-archived demand pool. Create one with the steps in §5.7.2, or un-archive the destination.
• No dashed red. Nothing has run, or the latest forecast sold everything.
• No gutter at all on a phone. #book-wires is hidden below the md breakpoint so the columns can stack. Rotate or widen to see the joiners.
• Headline above the board still talks. Even with no curves, the Rules headline still describes the cascade (how many supply shelves, how many demand baskets, whether supply outweighs demand). The joiners are the picture of named edges; the headline is the picture of volume.

5.7.2 How to create a solid joiner

A solid joiner is not a separate object. It is the picture of one fact: this supply shelf has a Shape match rule that names that demand basket. You create the fact on the supply card. The gutter then draws the cable.

You cannot paint a joiner from the Demand column. Demand Rules never offers SHAPE_MATCH, and the demand picker has no Target pool field. The dashed red spill curve is also not something you add: it appears only after a forecast leaves unsold supply (§5.6).

Before you start. You need at least one Supply pool and one Demand pool, neither ARCHIVED. You are on Book · Rules (…/strategy/pools). Widen the window so the gutter is visible (it is hidden on a phone).

Follow along. Seat: trader_1@webill.net (password supersecret). Path: T1 Renewables Pool → Strategy → Allocation → Rules. Pick a supply card whose Add a rule grid still shows SHAPE_MATCH — on T1 that is usually not Solar, because Solar already carries a shape-match. Flexible or Firm is the typical free shelf. You will point it at Industrial (or any named demand card).

1. On the Supply card that should own the cable, click Rules. The branded modal #rule-list-modal opens (Rules eyebrow, the pool name as the title, X at #rule-list-modal-close). Existing rules sit in an ordered list (#pool-rules-list).
2. Under Add a rule (#available-rules), click the SHAPE_MATCH card (#available-rule-SHAPE_MATCH). The card explains that this type pins the shelf to a buyer for chosen hours. The form heading becomes Configure Shape match. (If the grid is missing and you see All applicable rules are already on this pool at #no-available-rules, this shelf already has every supply type, including Shape match. Delete that SHAPE_MATCH row first, or pick a different supply card.)
3. TOU periods is a multi-select: Peak, Standard, Off-peak. These are TOU (time of use) windows on the South African tariff clock — pick the hours this dedication should apply. You can select more than one.
4. Target pool (edge, optional) is the field that creates the joiner. The prompt is Any. Any means the rule applies to the whole shelf and no curve is drawn. Open the list and choose a demand pool by name (Industrial, Commercial, Municipal, Trader Sales, …). That name is the basket the cable will point at.
5. Click Add rule. A new row appears in the list: type SHAPE_MATCH, a short parameter summary (the TOU periods you picked), and the usual ▲ / ▼ / Disable / trash controls. There is no extra flash — the list updating is the save.
6. Close the modal (X, Escape, or click outside). On a desktop-width board a solid curve now runs from that supply card to the demand card you named. Its colour is the supply type (Flexible is violet, Firm is dark slate, Wind is light blue, Solar is amber). The word on the curve is shape match.

If nothing appears. Check four things, in order: you named a Target pool rather than leaving Any; that demand card is not ARCHIVED; you are looking at a wide window (#book-wires is hidden on a stacked phone layout); you added SHAPE_MATCH on a supply card, not a demand card.

What the other supply types do. PRIORITY, PRO_RATA, MW_CAP and NOMINATION on the same picker have no Target pool field. Saving them does not draw a joiner. PRICE_OPTIMISE is demand-only in the picker, so you cannot start a price ≥ N c cable yourself; the T1 Wind price joiner is demo seed (§5.7.1).

After you save. Adding a rule is a structural change. The amber Forecast is behind this board banner returns until you press Run Forecast Now (§5.6). The dashed red rest → spill · N % only refreshes after that run.

How to take a joiner away. Open Rules on the same supply card. Trash the SHAPE_MATCH row. Closing the modal removes the curve. Disable greys the row in the list but does not hide the gutter curve today — delete the rule (or archive the destination pool) when you want the cable gone.

T1 on the demo — reading the amber, the blue, and the dashed red

Data. full_demo · seed 42 · T1 Renewables Pool (01a0ab54-e46b-7dde-b45f-73d119213c75) · Book · Rules.

On that seeded book the demo places a SHAPE_MATCH on Solar aimed at a demand basket, and a PRICE_OPTIMISE (floor 12.5 c/kWh) on Wind aimed at a demand basket. After a forecast that cannot sell everything, the gutter therefore typically shows:

The Rules picker on a supply card does not offer PRICE_OPTIMISE (that type is demand-only when you add a rule yourself). The T1 seed can still store a price-floor rule on Wind with a named buyer, and the gutter will draw it. Treat that as demo wiring you can delete; your own books only get a price joiner from a supply shelf if a rule with a target actually sits there.

Extra shelves (for example a Custom Approved Override pool) can sit in the cascade without adding a curve, as long as they have no destination rule.

What “done” looks like. You can point at each solid curve and name the supply card, the rule, and the demand card. You do not treat amber as a synonym for shape match. You know the dashed red is unsold forecast energy, and you know a phone-width window will hide the whole gutter.

5.8 Explanations — How to Read Rules

Explanations (#book-explanations-open) sits on the Rules toolbar next to Run Forecast Now. It opens a right-hand drawer (#book-explanations-drawer) — the same chrome as Overview Explanations (§4.11): green header, X top-right, backdrop, Escape.

Eyebrow: Book · Rules. Title: Explanations. Above the terms sits How to Read the View (#book-how-to-read) — collapsed until you press #book-how-to-read-toggle. It is the short in-product reading order for this mode: thread strip, issues rail, headline, Supply · Demand, outcome badges, wiring gutter, book rates, unassigned contracts.

The A–Z list is only the words on the Rules board (ACTIVE through Wiring gutter). It does not include Forecast tiles such as Margin@Risk, or Outcomes warehouse words. Open Forecast or Outcomes and the same control teaches that mode instead.

Close with the X (#book-explanations-drawer-close), the backdrop, or Escape.

Data. full_demo · seed 42 · stage S09 (settlement matrix) · last re-derived 2026-07-18 against the full_demo manifest.

Follow along. Seat: trader_1@webill.net (password supersecret). Path: T1 Renewables Pool → Strategy → Pools. You should see supply and demand structure before assigning or changing a member.

6. Book · Forecast — what the engine will do

You need an answer you can reproduce, not an editable dashboard number. A run is one immutable execution of the allocation engine for a scenario, grade and calendar month. Runs preserve the plan you had when you made a decision, while the index makes the current result easy to distinguish from its history.

Route: /org/trader-pools/:id/strategy/book?mode=forecast · Tab: Book · Mode: Forecast (the old /allocations route forwards here). Forecast is not a #strategy-nav tab — Allocation stays lit. Open it from the thread strip (#book-stage-forecast), a run link, or this URL.

Data. full_demo · seed 42 · stage S09 (settlement matrix) · last re-derived 2026-07-18 against the full_demo manifest.

6.1 The run picker, and what the mode shows

The Forecast mode reads one run at a time. The run picker under the headline is a month input and one chip per run of that month — Forecast P50, Adverse P90, Nominated, Actual — each with its status pill, a Simulations ×n count with any failures, Run backcast when the month is past and has a Forecast but no Actual, and New run. Pick a chip and everything below re-reads for that run; the URL carries &run=<id> so a link reproduces the view. Under the picker, Open the run ledger › drills to the run detail page (§8).

The mode then reads top to bottom. The table is a map. Each block is taught in full from §6.9.

There is no three-card “next moves” strip under Forecast. The thread strip already opens Nominations and Outcomes. Spill and shortfall that are this large get one Desk door each, the same way Rules offers Find offtake → Desk when leftover supply is above 20 % (§5).

The table below is what each run carries, and how to read it; on the picker the scenario is the chip, the period is the month input, the grade and status are the chip’s label and pill, and the four KPIs are the tiles.

The current forecast is the newest non-superseded COMPLETE run for that scenario, grade and period. A nominated run has its own current row; it does not replace your forecast simply because it was created later.

6.2 Row actions — Open and Run backcast

Two actions sit beside the run picker. They are not decoration: the picker is a triage list; these actions are how you prove a number or close the loop against meters.

Open →

Why it is there. KPI columns answer “how did the month look?” They cannot answer “which hour, which member, which rule?” The engine’s proof lives on the run detail page (§8). Open the run ledger › — under the picker and in the ledger fold — is the drill-through from the picker into that audit floor.

What it does. Navigates to /org/trader-pools/:id/strategy/allocations/<run id> for that immutable run — not “the latest for the month”, and not a live recalculation. You see:

• scenario, grade, status, period and the inputs hash (reproducibility fingerprint);
• a day-by-day stacked chart of supply vs demand;
• tabs for allocation / spill / shortfall / margin (and BESS when the run dispatched a battery);
• every line the engine wrote, with prices and applied rules.

Use Open → whenever a Coverage, Margin, Spill or Shortfall figure surprises you, when you need evidence for a dispute or nomination decision, or when you want to compare a superseded row with its successor side by side (open each in turn). Superseded and failed rows remain openable so history stays auditable.

Run backcast

Why it is there. Forecast and nominated runs plan with shapes and declarations. After the month settles, you need a parallel Actual run built from measured meter energy so you can score forecast accuracy, nomination mismatch and dashboard panels that wait on settled actuals. Actual is intentionally not a choice in the New run modal — meters must exist first. Run backcast appears in the picker for a past month that has a Forecast but no Actual yet.

What it does. On a row that qualifies, Run backcast:

1. Takes that row’s calendar month.
2. Re-runs the allocation engine with scenario ACTUAL (settled meter inputs instead of forecast shapes).
3. On success, adds a COMPLETE Actual row for the same period and refreshes forecast / nomination accuracy signals that depend on settled actuals.
4. If no member has settled meter rows for the month, nothing is written — you get a clear flash that settled data is not available yet.

When the control appears. Only on a Forecast row that is COMPLETE, and only while that period does not already have a non-superseded COMPLETE Actual. Nominated, Simulation and Actual rows do not show it; a forecast that already has its Actual for the month does not show it either (re-backcast is a deliberate later action once a newer Actual supersedes, or via ops schedules).

How to read the result. Filter to Actual (settled) or leave All and find the new ACTUAL row. Open → that Actual: the green Forecast vs settled actual panel (§8.2) shows Margin / Coverage / Spill / Shortfall Δ tiles, the month shape overlay (Forecast line vs Actual line), the same delivery period notice, and any Proxied members. That panel is the commercial forecast-error story; the day chart and tabs underneath then explain the hours.

Worked example — June 2026 backcast on T1 Renewables Pool

Data. full_demo · seed 42 · stage S07 (strategy historical runs) · T1 Renewables Pool · period June 2026 · grades P50. The Forecast and Actual KPI maps below are the sealed gallery pair for that month; shortfall hour lines are the seeded first-weekday slice on each run.

1. Starting point on the Runs index. You already have a completed Forecast for June. Meters for the month have settled, so Run backcast is visible on that row. After it completes, the index shows two current (non-superseded) companions for the same period:

2. Month roll-up — the numbers on each row. Coverage and residual shares are derived from the run KPI map with the same ratios the Runs header tooltips describe. Read them as snapshot dumps — the same card format as §4.10 · Snapshot the Overview actually shows.

Coverage      = allocated_mwh ÷ total_demand_mwh
Spill %       = spill_mwh ÷ total_supply_mwh
Shortfall %   = shortfall_mwh ÷ total_demand_mwh

MARGIN_R              = 331 500
MARGIN_R_PER_MWH      = 263
COVERAGE_PCT          = 81.3           ← 1 113 ÷ 1 369
SPILL_PCT             = 3.6            ← 44 ÷ 1 217
SHORTFALL_PCT         = 4.4            ← 60 ÷ 1 369
TOTAL_SUPPLY_MWH      = 1 217
TOTAL_DEMAND_MWH      = 1 369
ALLOCATED_MWH         = 1 113
SPILL_MWH             = 44
SHORTFALL_MWH         = 60

MARGIN_R              = 353 000
MARGIN_R_PER_MWH      = 266
COVERAGE_PCT          = 81.1           ← 1 126 ÷ 1 388
SPILL_PCT             = 3.9            ← 48 ÷ 1 234
SHORTFALL_PCT         = 4.7            ← 65 ÷ 1 388
TOTAL_SUPPLY_MWH      = 1 234
TOTAL_DEMAND_MWH      = 1 388
ALLOCATED_MWH         = 1 126
SPILL_MWH             = 48
SHORTFALL_MWH         = 65

MARGIN_R              = + 21 500       ← commercial headline improved
MARGIN_R_PER_MWH      = + 3
COVERAGE_PCT          = − 0.2 %        ← demand (+19) slightly outran allocation (+13)
SPILL_PCT             = + 0.3 %
SHORTFALL_PCT         = + 0.3 %
TOTAL_SUPPLY_MWH      = + 17           ← resource beat the P50 shape
TOTAL_DEMAND_MWH      = + 19           ← load ran above the profile
ALLOCATED_MWH         = + 13
SPILL_MWH             = + 4
SHORTFALL_MWH         = + 5

Forecast demand side : allocated + shortfall = 1 113 + 60 = 1 173
Forecast supply side : allocated + spill     = 1 113 + 44 = 1 157
Actual   demand side : allocated + shortfall = 1 126 + 65 = 1 191
Actual   supply side : allocated + spill     = 1 126 + 48 = 1 174

Pattern to expect on your own page: one Forecast row and one Actual row for the same period; Δ is Actual minus Forecast. A healthier margin with slightly worse spill/shortfall means prices and mix helped more than volume residuals hurt — confirm on the detail tabs before changing contracts.

3. Prove the residual on one hour (Open → → Shortfall tab). The seeded first weekday (1 Jun 2026) carries a shortfall line every hour. Three anchors show how Actual diverges from Forecast on the same clock:

00:00 gap growth      = 1 244 − 1 197 = + 47 kWh
12:00 gap growth      = 1 674 − 1 569 = + 105 kWh
18:00 gap growth      = 1 899 − 1 755 = + 144 kWh
Month SHORTFALL_MWH Δ = 65 − 60 = + 5 MWh   (= 5 000 kWh)

The evening hours dominate the month’s +5 MWh shortfall delta — that is why a nearly flat coverage % can still hide a procurement problem. On the Actual detail page, filter the Shortfall tab to 18:00 and note which demand membership owns the line; that membership is the candidate for firmer supply or a nomination change next month.

4. What you do next.

1. Keep the Forecast row as the plan you made; keep the Actual row as what meters proved.
2. If margin Δ is positive but shortfall Δ is also positive (as here), fix evening firmness before celebrating the rand outcome.
3. If you needed the opposite story (Actual weaker on margin), open Margin by TOU on both runs and find which period inverted.
4. Nomination Performance and forecast-error panels that were “awaiting settled actuals” can now light up for June — the backcast is their input.

Note. A scheduled monthly backcast also runs after settlement reconciliation (see the operations catalogue). Run backcast on the index is the on-demand control when you need the Actual lane before the calendar job, or for a month you have just finished settling.

6.3 Choose the scenario deliberately

New run opens the modal. The scenario chooses inputs, not a cosmetic label.

Use the Simulation Workbench for commercial candidates instead of manually creating a simulation run. It creates the required baseline and variant pair, keeps the candidate out of your live pool, and presents the deltas together.

6.4 Set period and grade

Inside New run, the Month picker covers a complete calendar month, hour by hour. Choose the month you intend to decide about; a high margin in August cannot justify a September nomination. Forecast grade chooses the supply confidence view (P50 expected or P90 adverse).

For a solar P50 interval of 10.00 MWh, the derived P90 is 10.00 × 0.94 = 9.40 MWh. For a wind interval of 10.00 MWh it is 10.00 × 0.86 = 8.60 MWh. That larger wind gap is why two books with the same P50 margin can carry materially different Margin@Risk.

Careful. P90 is an adverse planning view, not a guarantee that every interval will produce exactly 90% of P50. Use the pair to understand sensitivity, then inspect the allocation hours that cause the difference.

6.5 Worked New run — forecast August

Suppose you have adjusted a member in T1 Renewables Pool and need evidence now rather than at the next nightly cycle.

1. Open Book · Forecast and select New run in the run picker.
2. Choose FORECAST, choose August 2026 and retain P50 (expected).
3. Start the run. Its chip appears as pending, then running.
4. When it becomes COMPLETE, select Open → and confirm the pool, period and grade on the detail page.
5. Repeat with P90 (adverse) only if you need the risk pair; compare the two run-level margins rather than mixing their line items.
6. When meters for that month have settled, return to the Forecast row and select Run backcast to create the Actual companion; open the Actual for the Forecast-vs-settled panel (§8.2).

If the book has an active battery, its dispatch is included automatically using the two-pass method in §7.3. The completed detail page consequently includes the BESS evidence tab.

6.6 Understand superseding before you rerun

A successful newer run replaces the current row only within the same scenario, grade and period. The older row remains readable at reduced opacity so you can show what changed and when. Open → still works on superseded rows.

This is an audit feature. It prevents a manual rerun from rewriting yesterday’s decision evidence, while still ensuring one unambiguous current result.

6.7 Follow along — inspect, run, verify

Follow along. Seat: trader_1@webill.net (password supersecret). Path: Trader Pools → T1 Renewables Pool → Strategy → Runs. You should see newest-first rows with scenario, grade, status, KPI columns and, on the right, Open → (every row) and Run backcast on completed Forecast rows that do not yet have an Actual.

Filter to Forecast. Select Open → on the newest completed non-superseded row and skim the day chart and Shortfall tab, then return to the index. Create a P50 run for a month you are testing via New run. Wait for COMPLETE before comparing it with the prior row. If a completed Forecast shows Run backcast, use it only when you know meters for that month have settled — otherwise expect the “no settled meter data” flash. If a new run fails, read the error and keep using the previous completed result; no cleanup is required.

Runs answer “what did the engine decide?” The next chapter explains how the engine reaches those allocation decisions; §8 shows how to prove a particular result after you select Open →.

6.8 Explanations — How to Read Forecast

Explanations (#book-explanations-open) sits to the right of the Forecast headline. Same drawer shell as §5.8. Eyebrow: Book · Forecast.

How to Read the View is a different fold from Rules: this mode is one sealed run, not the wiring diagram. The reading order is thread strip, headline, run picker, five tiles, where the supply goes, where the margin is made and lost, long or short hour by hour, pool caps, battery fold, then Find offtake (only when this run spills more than 20 % of supply) or Cover shortfall (only when unmet demand is above 5 %).

The A–Z list is the words on Forecast (Actual through Where the supply goes). Wiring gutter is not here — that word lives on Rules. Warehouse rule is not here — that word lives on Outcomes.

The in-product drawer is a one-screen reminder. The rest of this chapter walks the same blocks with the T1 September numbers.

6.9 How to read Forecast, block by block

A dashboard number you can edit is a wish. Forecast is a sealed run — one pass of the engine for one scenario, one grade and one calendar month. You cannot type a new margin onto the tiles. You pick another run, or you start a new one.

This mode does not change the wiring. That is Book · Rules. It does not settle meters. That is Book · Outcomes. It answers: if this book ran for this month at this grade, where would the energy go and what would the money do?

Think of a bakery that already bought the day’s flour. Forecast is the bake sheet for one oven setting: how much bread sold, how much dough was leftover, and whether the till ended up or down. The recipe is Rules. The till tape after the shop closed is Outcomes.

Follow along. Seat: trader_1@webill.net (password supersecret). Path: Trader Pools → T1 Renewables Pool → Strategy → Allocation → Forecast on the thread strip (#book-stage-forecast). Deep link: /org/trader-pools/01a0ab54-e46b-7dde-b45f-73d119213c75/strategy/book?mode=forecast&month=2026-09&run=01a0b951-b50d-76c2-bfd8-840b7e. You should see the eyebrow Book · Forecast, a headline that names September P50, five tiles, then the two-by-two of charts and tables.

Data. full_demo · seed 42 · T1 Renewables Pool · run 01a0b951-b50d-76c2-bfd8-840b7ecc12cd · scenario FORECAST · grade P50 · period 1 Sep 2026 → 1 Oct 2026 · status COMPLETE · not superseded. Numbers below are that run’s KPI map and allocation lines, last read 2026-09-19. Compact Rand follows the product (R 13.1 m, R 786 k). Energy labels follow the product too (35.7 GWh, 979 MWh).

On screen, top to bottom

There is no #book-moves strip on Forecast. Nominations and Outcomes stay on the thread strip. The three ranked cards live on Outcomes only (§6A.14).

What “looks good” vs “looks bad” on this whole page. Good: coverage high, spill inside its cap, margin positive, caps read ok, no Find-offtake line. Bad — and this is T1 September — every demand hour is covered and almost all supply spills and the month loses tens of millions. Covered customers are not the same as a healthy book.

6.10 Thread strip, issues rail and headline

The problem. You can open Forecast from three doors (the strip, a run link, this URL) and lose the rest of the Book. The strip is how you stay on the same tab.

Thread strip (#book-thread). Three stages, left to right:

Forecast is ringed (data-on=true) on this mode. A negative margin prints rose; a positive one prints brand green. Clicking a stage navigates; it does not rerun the engine.

On this T1 September run the Forecast stage reads September P50 and −R 13.1 m, with pill spill 96.4 %. The note under it is 100 % covered · 96.4 % of supply unsold.

Where the strip is taught. The three-stage Book thread — §5. This section only names what Forecast lights and what the pill prefers.

Issues rail (#book-issues). Ranked cards under the strip, each with a colour dot (rose = bad, amber = warn, slate = info), a title, a body, and a door (Rules › / Forecast › / Outcomes ›). On T1 September the first card is Spill at 96.4 % of supply. The body lists the supply pools that broke their cap (Solar, Wind, Flexible, Firm) and ends Whatever the rules cannot sell, the forecast spills at the spill value. Next, Coverage outside its band on 2 demand pools — Commercial and Municipal sit at 100 % against narrower bands; the body adds Over-covered is over-bought. There is no shortfall card: unmet demand is 0 %.

Headline (#book-forecast-headline). One sentence from the run’s KPIs. The recipe, in order:

1. Empty KPI map: {Month} {Grade}: the run has no figures yet ({status}).
2. Spill > 50 % and margin negative (T1 September): {Month} {Grade}: every demand hour is covered, but {spill} % of supply finds no buyer and the month loses {abs margin}.
3. Shortfall > 20 %: unmet-demand sentence.
4. Margin negative (without that huge spill): loses / coverage / spill.
5. Spill > 20 % with a profit: earns but spills.
6. Otherwise: earns at coverage and spill.

T1 September hits rule 2:

September P50: every demand hour is covered, but 96.4 % of supply finds no buyer and the month loses R 13.1 m.

Lede (the grey line under the headline). Always: The engine’s month, one run at a time. Pick the run at the top; everything below re-reads. The same pool colours as the Rules board, so the bar that spills here is the card that breaches there.

Explanations sits to the right of that pair (§6.8).

6.11 The run picker and the doors around it

The problem. A month can hold a Forecast, an Adverse P90, a Nominated plan, several simulations and — once meters settle — an Actual. Mixing them on one canvas would lie. The run picker (#book-run-picker) makes you choose.

Month input. type=month, aria-label="Run month". Changing it patches ?month=YYYY-MM and reloads the chips. September cannot explain August.

Chips. At most four, in this order, and only when that slot has a run:

Each chip carries data-role="run-chip", data-scenario and data-status, plus a status pill: complete (green), pending / running (amber), failed (rose). The selected chip is ringed. Clicking it patches &run={id} so a shared URL reproduces the view.

September 2026 on T1 also shows a dashed Simulations ×n count (nine simulation rows in this dataset, including hypothetical candidates) and an amber n failed pill when any run failed. The count is not a chip — it does not re-read the tiles.

This URL pins a run. The walkthrough uses …&month=2026-09&run=01a0b951-b50d-76c2-bfd8-840b7e. That is the full-month P50 (01 Sep 2026 → 01 Oct 2026). The picker’s Forecast P50 slot can point at a different complete September forecast if a later one exists. Confirm the meta line under the picker before you quote a tile.

Meta line (data-role="run-meta"):

Run 01 Sep 2026 → 01 Oct 2026 · FORECAST · COMPLETE Open the run ledger ›

The link is /org/trader-pools/{id}/strategy/allocations/{run id} — the same door as §6.2 Open and §8.

New run (#book-new-run). Opens the branded modal #book-new-run-modal (header #book-new-run-header, title #book-new-run-modal-title, close #book-new-run-close or Escape or the backdrop #book-new-run-modal-backdrop). Fields: Scenario (FORECAST / NOMINATED / SIMULATION), Month, Forecast grade (P50 (expected) / P90 (adverse)). Helper: Actual (settled) runs are not started here; they are rebuilt from measured energy once meters settle. Run starts the worker; the button reads Running… while it is in flight. Scenario, period and grade are taught in §6.3–§6.5.

Run backcast (#book-run-backcast) appears only when the picker month is before today’s month, a Forecast exists, and no Actual exists yet. September 2026 on a 19 Sep 2026 clock is the current month — the button is hidden. How backcast works: §6.2.

6.12 The five tiles

The problem. A month of hourly lines is too long to scan. The five tiles (#book-forecast-tiles, each data-role="forecast-tile" with data-key) are the roll-up. Rose numbers are the leak. Green numbers are the ones that held.

They sit in one row from md up (two columns on a phone). Each tile is a label, a big value, and a note. They do not click through.

On this T1 September P50 run:

Tones: margin is rose when the month loses, green when it earns; coverage is amber under 70 %, else green; spill is rose over 20 %, else green; effective price is always slate; the battery tile is always green. The fifth tile is Take-or-pay (data-key=top) only when the run has no bess_arbitrage_margin_r key — see §6.17.

Margin · P50

Why it exists. This is the commercial headline: every allocation, spill and shortfall line of the month, netted.

The label carries the run’s grade so you do not mix a P50 margin with a P90 one. The value is compact Rand (R 13.1 m = about R 13,065,000). Full figure on the run: −R 13,064,676.

The note has two words:

• Contracted ratio — sold supply ÷ all supply. Here 3.6 %, printed as 4 %. Same story as 96.4 % spill: 3.6 + 96.4 = 100.
• Margin@Risk (MaR) — share of P50 margin lost in the P90 case. This run did not store mar_pct, so the tile prints MaR 0 %. That is a missing pair, not “no risk”. September’s Adverse P90 chip (when present) is a separate simulation; do not treat MaR 0 % as a clean bill.

Looks good: a green value, contracted ratio near the share you meant to sell, MaR filled from a real P90 companion. Looks bad: a rose tens-of-millions loss while coverage is 100 % — you paid for energy nobody took.

Coverage

Why it exists. Did the buyers get served? Coverage is allocated ÷ demand. Shortfall is the leftover demand (Shortfall %).

T1 September: 100 % coverage, 0 % shortfall, 2 demand pools with energy in the run (Commercial and Municipal). Industrial and Trader Sales sat idle — they do not count here.

Tone turns amber under 70 %. 100 % is green. That is not the same as a healthy book: you can cover two small baskets and still drown in spill.

Looks good: high coverage and shortfall inside the demand cards’ bands. Looks bad: a green 100 % that hides a missing buyer class, or a high average that still leaves evening gaps (open §6.15 before you celebrate).

Spill

Why it exists. How much flour never became a loaf? Spill is unsold supply ÷ all supply (Spill %).

T1 September: 96.4 % of 59,600 MWh supply was leftover — 57,437 MWh. The note prices that leftover at the book’s spill value: −R 16.5 m. That is the spill lines’ margin (buy price minus the dump price), not the month’s net. The month’s net is the Margin tile; spill is the leak inside it.

Tone is rose over 20 % — the same threshold that shows Find offtake (§6.19).

Looks good: spill inside each supply card’s cap (15 % Solar, 12 % Wind, 20 % Flexible, 5 % Firm — §5.1). Looks bad: 96.4 %. You cannot nominate your way out of a book that bought ~24 times the demand it can serve.

Effective price

Why it exists. Two prices live on one tile because “what did we earn?” has two honest answers.

• The big number is effective price in c/kWh on sold energy only. T1: 157 c — about R 1,570 / MWh on the 2,163 MWh that found a buyer.
• The note’s realised R/MWh divides the same sold margin by all generation, unsold hours at zero. T1: R 57 / MWh. Check: 1,570 × 3.6 % ≈ 57. The gap is the spill.
• HHI supply is concentration of supply members (0 = many equal sources, 1 = one source). T1: 0.41 — a few members do most of the volume, not a monopoly.

The tile is slate on purpose: a high cents figure on a thin sold slice can look healthy while the month loses R 13.1 m.

Looks good: realised and effective sit close (little unsold energy) and HHI is not a single-member spike. Looks bad: 157 c next to R 57 / MWh.

Battery — or Take-or-pay

T1’s book seats Rosslyn BESS — 5 MW / 20 MWh on Firm. The run stored battery KPIs, so the fifth tile is Battery, not Take-or-pay. The value R 0 and the note 0.00 cycles/day · absorbs 0.0 % of spill mean the battery was present and did not work this month — it did not soak the 57 GWh leftover.

ToP utilisation for this run is 2.5 % (almost none of the Firm take-or-pay volume earned). That figure is not on a Forecast tile while the battery key exists. When a book has no battery KPIs, the fifth tile reads Take-or-pay and that utilisation is the value.

Looks good (battery): a positive arbitrage value and a real cycles/day number. Looks bad: R 0 beside 96.4 % spill — the battery did not save the bake.

6.13 Where the supply goes

The problem. The Spill tile says how much leftover you have. It does not say which shelf leftover it, or which basket bought the slice that sold.

On screen: #book-flow. Title: Where the supply goes. Intro on the card:

Each supply pool’s energy for the run, split into what the cascade sold to the demand pools and what it spilled. The demand pools are the same cards, in the same order, as on the Rules board.

This is a server-drawn SVG (no hover, no separate legend). Colours are the legend: they match the Rules cards.

How to read it, left to right.

1. Left stack. One bar per supply pool that moved energy, tallest first. The label is {name} {energy}.
2. Ribbons. A coloured band from a supply bar to Sold is that pool’s allocated energy. An amber wash from the same bar to Spill is that pool’s leftover.
3. Right stack. Demand bars in Rules priority order (not by size). T1 only draws Commercial and Municipal — Industrial and Trader Sales had no demand in the run.
4. Right labels. Sold {energy} · {sold %} %. Under it, in rose: Spill {energy} · {spill %} % and {rand} at the spill value. A Shortfall {energy} line appears only when unmet demand is above zero — T1 has none.

Empty copy, when the run has no allocation lines: No allocation lines in this run.

Where the Overview cousin is taught. Overview §4.4 is a ribbon map of the snapshot run, coloured by margin, with hover. This Forecast chart is the picked run, coloured by pool type, with no hover. Same marketplace, different job.

Worked reading — T1 September P50

Conservation first. The chart must add up before you argue a ribbon.

Supply (alloc + spill)     = 59,600 MWh
Sold (allocated)           =  2,163 MWh   →  3.6 %   → label “Sold 2.2 GWh · 3.6 %”
Spill                      = 57,437 MWh   → 96.4 %   → label “Spill 57.4 GWh · 96.4 %”
Shortfall                  =      0 MWh
Demand (sold + shortfall)  =  2,163 MWh
Sold + spill               = 59,600 MWh = supply ✓

Left stack, tallest first:

Flexible has no sold ribbon — the whole bar washes into Spill. Approved Override Supply is ACTIVE on Rules and sent nothing this run, so it is absent.

Where the 2.2 GWh actually landed (the five allocation legs):

Demand bars, Rules order: Commercial 979 MWh, Municipal 1.2 GWh. No shortfall notch.

Looks good: fat coloured ribbons into Sold, a thin amber spill outlet, every live demand bar present. Looks bad: a left stack that is almost entirely amber wash — T1.

Where to go next. A fat leftover from one shelf: open that card on Rules and the joiners in the wiring gutter. Need a buyer for the leftover: Find offtake → Desk (§9). Need the hour: §6.15 then the ledger (§6.18).

6.14 Where the margin is made and lost

The problem. The Margin tile is one number. You still need to know whether you lost money on prices (sold energy was cheap) or on leftover dough (sold energy was fine; unsold energy sank the month).

On screen: #book-margin. Title: Where the margin is made and lost. Intro:

By time of use and by source pool. Sold energy earns; spilled energy loses buy price minus the spill value.

Two small tables, then one verdict sentence.

By TOU (data-role="margin-tou"). Rows in clock order: Peak, Standard, Off-peak, then Battery / untimed when a line has no TOU. Positive margins print green with a leading +; negative ones print rose with −. Empty body: No margin breakdown in this run.

By source (data-role="margin-source"). One row per supply pool that sold, with the pool colour chip, the name, and the word sold. Sorted richest sold-margin first. A final Spill · all pools row (data-role="margin-spill") is the amber chip and the leftover’s rand. Flexible does not appear on T1 — it sold nothing.

The verdict under the tables (margin_note):

T1 September is the first sentence. Remember: the product says “every hour” while the test is every source pool’s sold margin. The TOU table is allowed to be red — it includes spill.

Where the Overview cousin is taught. Overview §4.8 is a month × hour heatmap of the snapshot. This card is two tables for the picked run.

Worked reading — T1 September P50

By TOU (these rows include spill):

Off-peak is the worst clock band because that is when the book is longest and the leftover is cheapest to dump. The +R 12 k is BESS reserve lines, not a dispatch profit.

By source (sold legs only) plus the spill row:

Sold sources          =  2.5 m + 786 k + 94 k  ≈ +R 3.4 m
Spill                 =                         −R 16.5 m
BESS reserve (untimed)=                         +R 12 k
Month (Margin tile)   =                         −R 13.1 m
3.4 m − 16.5 m + 0.012 m ≈ −13.1 m ✓

Looks good: green TOU rows, green source rows, a small or missing spill row. Looks bad: green sold rows and a spill row that is five times larger — a wiring-and-demand problem. Rewire or sell offtake; do not first cut the sell price.

Where to go next. Price problem → member buy/sell on Rules. Wiring-and-demand problem → gutter and Find offtake. Hour proof → §8 Margin tab.

6.15 Long or short, hour by hour

The problem. A month total can hide a noon glut and an evening gap. You need the shape of a typical day.

On screen: #book-day. Title: Long or short, hour by hour. Intro:

A typical day of the run: supply stacked by pool against the demand line. Above the line is long (spill), below it short (shortfall). The Overview’s clock, now per pool.

A typical day is every line of the run grouped by SAST clock hour (the stored UTC hour + 2), then divided by the number of days in the period. This run is 1 Sep → 1 Oct, so each point is the month’s that-hour total ÷ 30.

How to read the SVG (no hover, no clip strip, no Tomorrow toggle — those belong to Overview §4.6):

1. X axis. 0h … 23h SAST, labelled every four hours (0h, 4h, 8h, 12h, 16h, 20h).
2. Y axis. Energy. Ticks use the same compact units as the flow chart (kWh / MWh / GWh). On T1 the noon stack is about 136 MWh, so the axis is in MWh.
3. Stacked areas. One band per supply pool, coloured like Rules. The engine draws the smallest shelf at the bottom and the largest on top: Flexible, Firm, Solar, then Wind.
4. Solid ink line = demand that hour (allocated delivered + shortfall). The word demand sits at 23h.
5. Long vs short. Stack above the line = long (you will spill). Line above the stack = short (you will buy top-up).

Empty copy: No hourly lines in this run.

Worked reading — T1 September, three clock hours

Every hour is long. Demand is a thin ink thread under a tall stack.

Looks good: the stack follows the demand line; a little amber only at solar noon if you are slightly long; a rose evening only if you sold firm hours you cannot cover. Looks bad: a stack that never touches the line — T1. You are not “a bit long at noon”; you are long for 24 hours.

Where to go next. Noon leftover → sell daytime offtake on Desk, or stop seating more solar until the gutter has a buyer. An evening short (not present here) → firm shelf or Cover shortfall. Hour proof → the ledger.

6.16 Pool caps, this run

The problem. Rules shows a badge on each card. Forecast repeats those badges as a table for this run, so you can quote them next to the tiles.

On screen: #book-caps. Title: Pool caps, this run. Intro: The Rules board’s outcome badges, as a table.

Columns: Pool, Result, Cap / band, and a pill.

A supply row is {colour chip}{name} spill. Result is the run’s spill %. Cap / band prints ≤ {cap} %. A demand row is {chip}{name} coverage. Result is coverage %. Cap / band prints {min}–{max} % when both ends exist.

Only ACTIVE pools that actually took part (active?) appear. Drafts, archives, and live pools with no energy in the run are omitted. Empty table: No pool took part in this run.

Worked reading — T1 September P50

(97.5 % Firm spill rounds to 98 % on the table.) Industrial and Trader Sales have bands on Rules and no demand in this run, so they are absent. The four breach rows are the same four names on the issues rail. The two outside rows are the over-covered baskets: every hour was served, which is more cover than those cards asked for.

Looks good: every pill ok. Looks bad: a column of breach plus outside — you both dumped most of the flour and over-fed the two baskets that showed up.

Where to go next. Press Rules › on the issues card, or open the named pool on the board. Do not edit a cap to hide a 96 % spill; seat a buyer or take less supply.

6.17 Battery dispatch

The problem. A battery can soak leftover noon energy and give it back in the evening. You need to see whether this run used it.

When the fold exists. #book-battery renders only when the run’s KPI map has bess_arbitrage_margin_r — even if the value is zero. It is a <details> that opens by default. Summary:

Battery dispatch · {member label} · from this run shown only for books with an active battery

T1’s member label is Rosslyn BESS — 5 MW / 20 MWh.

Five mini tiles (data-role="battery-tile"):

Under the tiles: The dispatched day, the hourly schedule, the rejected-pairs trust device and the asset settings live on the full dispatch page. Open full dispatch › goes to /org/trader-pools/{id}/strategy/bess. The link does not carry ?run= — the destination page loads the book’s latest complete Forecast that already has battery intervals (§11.11).

Where this is taught. This section is the Forecast fold — five tiles and a door. The full /strategy/bess page (header, which run it opens, facts, chart, moves, hour table, empty states) is §11.11. How the engine decides charge, hold and discharge is §11.2.

When the fold is absent. A book with no battery KPIs does not render #book-battery. The fifth tile then becomes Take-or-pay. That is honest: there is nothing to fold.

T1’s honest empty-ish state. The fold is there. The battery did not dispatch. One BESS_CHARGE line (2.3 MWh) and +R 12 k of BESS_RESERVE sit on the run; they do not move the arbitrage tile off R 0 and they do not absorb the 57 GWh spill.

Looks good: cycles above zero, spill absorbed above zero, or firm hours backed above zero. Looks bad: a named BESS and a row of zeros next to a 96 % spill tile.

6.18 The run’s line ledger

The problem. Charts argue. Lines prove.

On screen: #book-ledger-fold — a dashed fold, closed until you open it. Summary:

The run’s line ledger · Allocations · Spill · Shortfall · Margin · Audit one SAST day at a time

The body: Every allocation, spill and shortfall line of this run with its provenance, priced hour by hour. Then Open the run ledger › — the same allocations URL as the meta line under the picker. The destination heading has ← Pool · Forecast (#allocation-back-to-forecast) back to this Forecast for the same pool, month, and run.

This fold does not inline the day table. It is a door.

Where this is taught. The destination /org/trader-pools/:id/strategy/allocations/:run_id — header, ← Pool · Forecast, day picker, every tab, empty states, and the T1 September walk — is §8. Start at §8.6 for the click path.

Use it when a tile surprises you, when you need a line for a dispute, or when you want to compare this run with a superseded companion.

6.19 Find offtake, Cover shortfall, and empty states

The problem. After you have read the leak you still need one obvious next click. Forecast does not invent a third “moves” gallery.

Find offtake (#book-forecast-offtake-cta) is a single line under the ledger, only when this run’s spill is above 20 %:

Spill has no buyer in this run. Find offtake → Desk

The link (#book-forecast-offtake-desk) opens Simulations (/strategy/simulations) — §9. T1 September is 96.4 %, so the line is present.

Cover shortfall (#book-forecast-cover-cta) is the twin, only when unmet demand is above 5 %:

Demand is served at the top-up price. Cover shortfall → Desk

T1 September is 0 % shortfall, so the line is absent. Both lines can appear together on a book that is long in one hour and short in another.

Nominations and Outcomes are not repeated here. They stay on the thread strip.

Empty and partial states you will actually see

What to do next on T1 September

1. Believe the headline: the customers who showed up were served; the month still loses R 13.1 m on leftover supply.
2. Open Rules and sit a buyer on the gutter, or press Find offtake → Desk and price a sell-side candidate.
3. Do not start from the 157 c effective price — that is the thin sold slice.
4. When you need a line, open the run ledger. When the month has settled, read Outcomes — Forecast will not rewrite itself into a warehouse.

6A. Book · Outcomes — what actually settled

Route: /org/trader-pools/:id/strategy/book?mode=outcomes&month=YYYY-MM · Tab: Book (#strategy-tab-book) · Mode: Outcomes (#book-stage-outcomes on the thread strip; BookLive). The old …/book?month= link and the old …/chains route open here.

A trader buys from several sources and sells to several sinks in the same month. Forecast (§6) answers “what would the engine do?” Outcomes answers “what actually cleared?” — every buy leg and sell leg on which this pool sat, the warehouse rule that you cannot have sold more than you procured, the chains that make the month up, and how far the forecast missed the ledger.

Think of Book as a warehouse stocktake. Forecast is the bake sheet you wrote before the month. Outcomes is the till tape after the shop closed: who delivered flour, who left with bread, and whether you sold more loaves than you ever took in.

6A.1 The trader’s problem

Opening one PPA’s settlement detail shows one leg. Your commercial question is the book: who gave me energy, who did I give it to, does Σ sell ≤ Σ procured still hold, and did the plan I ran a month ago look anything like this?

6A.2 What you see

The table is a map. Each block is taught in full from §6A.5.

Data is assembled by Statement.TraderBook.build/3 and ChainExplorer — the same projections statements use. No settlement maths is redone; line amounts and period totals are read off the ledger.

Follow along. Seat: trader_1@webill.net (password supersecret). Path: Trader Pools → T1 Renewables Pool → Strategy → Allocation → Outcomes. Deep link: /org/trader-pools/01a0ab54-e46b-7dde-b45f-73d119213c75/strategy/book?mode=outcomes&month=2026-09. You should see the thread strip, the headline, the two bars and the book check, the money ledger, the buy/sell panels, the chains and the three move cards.

6A.3 How to read it

1. Read the badge before you argue a month commercially. A red badge means a sell leg has no supply behind it — a paper commitment, or supply legs missing from settlement.
2. Buy legs are “who gave me what”; sell legs are “who I gave it to”.
3. The upper bar is the most that can honestly be sold. Sold energy past it is the red segment, and the same energy shows as uncovered out-legs in the book check.
4. Net margin is your seat’s P&L (sell revenue − buy cost − network you bore). Counterparty prices never appear.
5. The drift panel is the reconciliation: if the forecast prices thirty times the supply that settles, fix the wiring on the Rules board before trusting the forecast money.

6A.4 Explanations — How to Read Outcomes

Explanations (#book-explanations-open) sits to the right of the Outcomes headline. Same drawer shell as §5.8. Eyebrow: Book · Outcomes.

How to Read the View is the warehouse order: thread strip, headline, month picker, procured against sold, the month’s money, buy legs · sell legs, chains, forecast said versus book settled.

The A–Z list is the words on Outcomes (Balanced through Warehouse rule). It does not repeat the Rules gutter or the Forecast flow chart.

The walk below is that same order, with every control and the T1 numbers.

6A.5 How to read Outcomes, block by block

A dashboard number you can edit is a wish. Outcomes is a warehouse stocktake for one settled month. You cannot type a new net onto the money card. You pick another month, or you open a chain and prove a leg.

This mode does not change the wiring. That is Book · Rules. It does not run the engine. That is Book · Forecast. It answers: who gave me energy, who did I give it to, and did I sell more than I bought?

Follow along. Seat: trader_1@webill.net (password supersecret). Path: Trader Pools → T1 Renewables Pool → Strategy → Allocation → Outcomes on the thread strip (#book-stage-outcomes). Deep link: /org/trader-pools/01a0ab54-e46b-7dde-b45f-73d119213c75/strategy/book?mode=outcomes&month=2026-09. You should see the eyebrow Book · Outcomes, a headline that names September’s 7.2 GWh sale, then the conservation chart beside the money card.

Data. full_demo · seed 42 · T1 Renewables Pool (01a0ab54-e46b-7dde-b45f-73d119213c75) · last read 2026-09-19 from TraderBook.build/3 and settlement_periods on wb_wheeling_dev. Compact Rand follows the product (−R 469 k, R 12.9 m). Energy labels follow the product too (460 MWh, 7.2 GWh). The live Outcomes URL was not opened in a browser this pass — Phoenix was up; the login door blocked automation. The numbers are the same projections the page renders.

On screen, top to bottom

#book-moves is Outcomes-only. Forecast has no three-card strip (§6.9). Rules has none either. Nominations stay off this page — they are not a #strategy-nav tab. The three ranked cards stay here so a broken warehouse has an obvious next click.

What “looks good” vs “looks bad” on this whole page. Good: green Σ sell ≤ procured, chains balanced, forecast and settled net sit close, the issues rail is quiet on this month. Bad — and this is T1 September — the lower bar runs 15.7× past the upper one, three sell legs sit outside any linear chain, and the month loses R 469 k on a 7.2 GWh paper sale.

6A.6 Route, Book tab, thread strip and issues rail

The problem. You can open Outcomes from four doors and lose the rest of the Book. The strip is how you stay on the same tab.

#strategy-nav. Overview · Allocation · Desk · Dashboard. Allocation is lit (#strategy-tab-book.tab-active). Outcomes is not a fourth room tab. There is no #strategy-tab-book-outcomes and no #book-mode-switch on this page. Book stays the tab; the thread strip switches the mode.

How you get here.

The default month and the strip’s month are not the same clock. The page without ?month= lands on today minus one month. The strip names the last month that has any received or supplied energy, walking back at most six months from today. On T1 that last month is September, even if you then pick August in the chips.

Thread strip (#book-thread). Three stages, left to right:

Outcomes is ringed (data-on=true) on this mode. A negative net prints rose; a positive one prints brand green. Clicking a stage navigates; it does not re-settle a period.

On T1, 19 Sep 2026, the Outcomes stage reads September · −R 469 k with pill not balanced. The note under it is 7.2 GWh sold against 460 MWh bought · 3 sell legs with no supply behind them.

That is not the older August · −R 381 k · not balanced line. August is a different warehouse: 186 MWh bought, nothing sold, the warehouse rule holds, net −R 196 k. Do not paste an Overview figure onto this strip. Confirm the month word and the pill before you quote the net.

Where the strip is taught. The three-stage Book thread — §5. This section only names what Outcomes lights and which month it prefers.

Issues rail (#book-issues). Ranked cards under the strip, each with a colour dot (rose = bad, amber = warn, slate = info), a title, a body, and a door (Rules › / Forecast › / Outcomes ›). The September-unbalanced card is the Outcomes one:

September book does not balance 7.2 GWh sold against 460 MWh bought. 3 out-legs uncovered. Outcomes ›

T1 also carries the Forecast/Rules cards the other modes already taught: spill around 96 % of supply, coverage outside its band on Commercial and Municipal, and a draft pool with no members. Those doors open Rules or Forecast. They do not change the warehouse.

When no month in the last six has energy, the rail adds Nothing has settled yet and points at Outcomes.

6A.7 Headline, lede and Explanations

The problem. The charts are true and still easy to argue past. The headline is the one sentence the ledger wants you to believe first.

Eyebrow. Small brand-green caps: Book · Outcomes. Same pattern as Forecast’s Book · Forecast.

Headline (#book-outcomes-headline). One sentence from the month’s statement and book check. The recipe, in order:

1. No statement at all: {Month Year} has no settled legs yet.
2. Statement exists, but both sides are zero: {Month} settled nothing: no buy leg and no sell leg reached FINAL.
3. Book check is not balanced (T1 September): {Month} settled {bought} bought and {sold} sold — the book does not balance: {N} sell leg(s) have no supply behind them, and the month lost (or made) {abs net}.
4. Something bought, nothing sold (T1 August): {Month} settled {bought} bought and nothing sold — the book carried {abs net} of supply cost with no revenue against it.
5. Otherwise: bought, sold, balanced, and a net loss or margin.

T1 September hits rule 3:

September settled 460 MWh bought and 7.2 GWh sold — the book does not balance: 3 sell legs have no supply behind them, and the month lost R 469 k.

T1 August hits rule 4:

August settled 186 MWh bought and nothing sold — the book carried R 196 k of supply cost with no revenue against it.

T1 June hits rule 2 (a statement with empty legs):

June settled nothing: no buy leg and no sell leg reached FINAL.

Lede (the grey line under the headline). Always: What the ledger says happened, one settled month at a time. The statement first, the chains that make it up underneath, and the forecast that predicted it beside it. A month that does not conserve energy is flagged before any money is read.

Explanations (#book-explanations-open) sits to the right of that pair. The drawer, the How to Read fold, and the A–Z list are §6A.4. Eyebrow inside the drawer: Book · Outcomes. Close with #book-explanations-drawer-close, the backdrop, or Escape. The walk you are reading is the How to Read order written out.

Outcomes does not open the Forecast New run modal. There is no #book-new-run on this mode.

6A.8 The month picker

The problem. A book has several closed months. Mixing them on one canvas would lie. The picker makes you choose.

Arrows. Ghost chevrons (phx-click="shift_month", by="-1" / by="1"). Each click patches ?month= one calendar month and clears any selected chain.

Month input. type=month, aria-label="Settlement month", phx-change="select_month". Changing it patches ?month=YYYY-MM. September cannot explain August.

Chips (#book-month-picker). The last six calendar months from today, newest first, each a patch link:

…/book?mode=outcomes&month=YYYY-MM

A month is settled on a chip when the book check sees received kWh or supplied kWh above zero. Status FINAL is not required. T1 September is still DRAFT on every September period — the chip still reads not balanced, because the energy is there.

On a 19 Sep 2026 clock the six chips are Sep, Aug, Jul, Jun, May, Apr:

Recomputed stamp ([data-role=chains-computed-at]). When the chain index has a clock: recomputed {dd Mon HH:MM}. Honest “as of now”, not an issued report. Empty when the index has no timestamp.

The URL also keeps &chain= and &unbalanced=1 when those filters are on. Changing the month drops the chain. Toggling the filter keeps the month.

6A.9 Procured against sold — the conservation chart

The problem. Two tables of counterparties do not show, at a glance, whether you sold more than you bought. The chart is the warehouse door.

Card (#book-conservation). Title Procured against sold. Helper: Energy conservation for the month. In-legs are what the book received from its suppliers; out-legs what it delivered to buyers. Sold cannot exceed bought unless a sell leg is a paper commitment with no chain behind it.

How to read the SVG (role="img", Procured against sold energy):

1. Bought · in — the upper green bar. Length is procured kWh (delivered on buy legs). The number to the right is the compact label (186 MWh, 460 MWh).
2. Sold · out — the lower demand-coloured bar. Length is sold kWh (supply on sell legs), clipped to what was bought.
3. Red tail — only when sold > procured. It starts at the bought ceiling and runs to the sold tip. The label adds · {energy} with nothing behind it.
4. Dashed vertical — drops from the bought tip. Caption: what was bought · the most that can honestly be sold.
5. Sentence under the bars. Over: A balanced month has the lower bar no longer than the upper one. This month is {ratio}× over. Inside: Sold energy sits inside what was bought. Empty: Nothing sold this month.

T1 August — nothing sold, so no red tail:

Bought · in → 186 MWh Sold · out → 0 kWh Sentence: Nothing sold this month. (the sold bar is empty; the ceiling still marks 186 MWh.)

T1 September — the picture you open the page to see:

Bought · in → 460 MWh Sold · out → 7.2 GWh · 6.7 GWh with nothing behind it Sentence: A balanced month has the lower bar no longer than the upper one. This month is 15.7× over.

7.2 GWh is 7,200,000 kWh on one OffTaker A period. 460 MWh is 459,740 kWh delivered on the IPP3 buy legs. 7,200,000 ÷ 459,740 ≈ 15.7. The red tail is the paper sale.

Book check ([data-role=book-check]), under the SVG:

Book check · {fan-out · }{N} in-legs · {N} out-legs · {N} uncovered

August: 2 in-legs · 0 out-legs · 0 uncovered and green Σ sell ≤ procured.

September: 5 in-legs · 4 out-legs · 3 uncovered and rose Sold more than procured. The chain walk keeps one next hop; the other three September sells never join a linear path. Book still owns them. That is why the statement and the chain list can disagree on count — §6B.8.

Empty card. When there is no statement, the two-column grid is replaced by a dashed panel: No settled statement for {Month Year}. June on T1 still has a statement (empty legs), so you get the chart at zero rather than this dashed empty. The dashed empty is the “Ash could not build a statement” state.

6A.10 The month’s money

The problem. Energy conservation is not a P&L. You can pass the warehouse test and still lose money — or fail it and still invoice a sliver.

Card (#book-money). Title The month’s money. Helper: From what was bought to what was kept, own seat only.

Five rows ([data-role=money-row], data-key on each). Costs print with a minus; revenue with a plus. Gross and net are extrabold. The net cell is [data-role=net-margin].

Exact cents behind the compact labels, August: buy 15,841,730 · network 3,727,466 · sell 0 · net −19,569,196. September: buy 39,077,875 · network 9,194,793 · sell 1,384,933 · net −46,887,735.

Imbalance, curtailment credits, trader-margin lines and levies sit on the ledger. They do not appear on this card. The card is the seat’s energy-and-network story, not a restatement of every category.

Forecast note under the list, when the drift series has a P50 for a month: The forecast for {Month} said {P50} at P50; the ledger settled {net}. See the drift panel. On T1 September that is The forecast for September said −R 12.9 m at P50; the ledger settled −R 469 k. The plan priced leftover supply the warehouse never bought.

Counterparty prices never appear. You see your cost, your network, your revenue.

6A.11 Procured (buy legs) and Sold (sell legs)

The problem. The bars collapse counterparties. You still need the names.

Two tables, side by side from lg up.

Procured (buy legs) (#trader-book-buy). Columns Source and kWh. One row per buy period ([data-role=buy-leg]). Footer Σ buy ([data-role=procured-kwh]) when the list is not empty. Empty body: No settled buy legs this month.

kWh is a whole number, no compact MWh label — 186373, not 186 MWh. The compact label lives on the conservation chart.

T1 August

T1 September

IPP1’s DEMO-SUPPLY period is a live buy PPA that delivered nothing. It still counts as an in-leg. IPP3 is the thread-cast generator (IPP3 Midrand Hybrid). Several September rows share one agreement because settlement stored more than one period start in the month.

Sold (sell legs) (#trader-book-sell). Columns Sink and kWh. Rows [data-role=sell-leg]. Footer Σ sell ([data-role=sold-kwh]). Empty body: No settled sell legs this month — that is August.

T1 September

OffTaker A Rosebank Mall is the thread-cast sink. 7,200,000 kWh on one period is the paper sale the red tail is made of. Delivered on that same period is only 16,293 kWh — the table shows supply kWh on sell legs, not delivered. That is deliberate: the warehouse asks what you committed, not what the customer took.

There is no sort, no filter, and no row click on these tables. To study a path, use Chains.

6A.12 Chains — pick a path

The problem. The statement is the whole warehouse. A break usually lives on one route. Chains is the map of those routes.

Where this is taught. Every card on an opened chain — rail, volume bridge, money, legs, book-check strip — is §6B. This section is the Outcomes chrome around that detail: the list, the filter, the click, and what T1 shows.

Layout (#trader-chains). List on the left (#chains-list, four columns from xl). Detail on the right (#chains-detail, eight columns) when a chain is selected. When none is selected: a dashed panel, Select a chain to study the volume bridge and margin.

List header. Chains · {total} ([data-role=chains-verdict]). When unbalanced > 0, a rose pill {N} unbalanced ([data-role=chains-unbalanced-count]).

Unbalanced only (#chains-unbalanced-toggle). Ghost border when off; rose btn-error when on. Click patches &unbalanced=1 (or drops it). The list then hides balanced paths. The URL is shareable.

Empty list (#chains-empty). No chains for this month — or No chains for this month (unbalanced filter on) when the toggle hid everything. August on T1 is the first form: buy-only months have no linear IPP → you → sink path of two or more legs. Turn the filter on in June and you get the second form (the LiveView test asserts that copy).

A row. Button #chain-row-{head_period_id}, data-role="chain-row", data-balanced="true|false". Body:

• Path label — Generator → You → Off-taker (T1 September: IPP3 Midrand Hybrid → You → OffTaker A Rosebank Mall)
• {N} legs
• Badge [data-role=chain-balance-badge]: green balanced or rose BREAK

Click patches &chain={head_period_id}. The selected row is lightly highlighted. A deep link may pass any leg’s period id; the screen still opens the whole chain.

What opens on the right (#chains-detail → <.chain_detail>):

T1 September should open the IPP3 → You → OffTaker A path. Expect a BREAK or a loud unexplained residual: the sell supply is 7.2 GWh and the buy delivered is hundreds of megawatt-hours. That is the same 15.7× the conservation chart already shouted.

August has no chain to click. The dashed Select a chain… panel stays up.

6A.13 Forecast said, book settled

The problem. A plan that prices energy the ledger never saw will keep sending you to Desk. The drift panel is the six-month argument.

Card (#book-drift). Title Forecast said, book settled. Helper: P50 margin at forecast time against the settled net margin, last six months. The gap is forecast error you are paying for. This is the trader cousin of IPP Allocation’s Forecast vs settled card — same idea, different ids and a book-level net rather than a buyer carve.

How to read the SVG (Forecast margin against settled margin by month):

1. X axis — six month ticks (Mar … Sep when September is selected), ending on the picker month.
2. Y axis — compact Rand, including below zero.
3. Dashed grey line — forecast P50 (forecast_r). A twelve-month snapshot is scaled to one month (margin × 30 / days when the run is longer than 45 days) before it is plotted.
4. Solid green line + dots — settled net (settled_r), only for months that have procured or sold energy.
5. Legend (top right): dashed · forecast P50   solid · settled net.

Verdict ([data-role=drift-verdict]). No pair: No month has both a forecast snapshot and a settled statement yet. Otherwise: The forecast misses the ledger by about {average absolute gap} a month across {N} month(s). Fix the wiring before trusting the forecast money.

T1 September has a forecast snapshot around −R 12.9 m (the same order as the Forecast headline’s −R 13.1 m / −R 12.2 m) and a settled net of −R 469 k. The gap is about R 12.5 m. The plan assumed leftover supply would keep losing money at forecast scale; the ledger only bought 460 MWh and then booked a 7.2 GWh paper sale. Neither number is the one you want — and they are not the same number.

Empty SVG. When no month in the window has a snapshot or a statement: No forecast snapshots or settled statements in the last six months yet.

Pick August and the window is March–August. September’s snapshot drops off the right. July and August then have settled nets (about −R 196 k each) and usually no matching P50 snapshot — the verdict falls back to No month has both… until a snapshot lands in that window.

6A.14 The three moves — Outcomes only

The problem. After you have read the warehouse you still need one obvious next click. Forecast solved that with a single Desk line. Outcomes keeps a three-card strip.

#book-moves. Three columns from lg. Each card is [data-role=book-move] with data-key. The first card has the brand border. Open › is the door. The worth figure sits on the right.

This strip does not render on Forecast or Rules. Do not look for it under the Forecast ledger. Do not remove it from Outcomes to “match” Forecast.

cover_sale is omitted when the book check is missing or the month is balanced. August on T1 therefore shows two cards (Reconcile, Dashboard) in the three-column grid. September shows all three. The first September card is the cover-or-cancel one, worth R 469 k.

There is no fourth card for Nominations. Nominations is not a strategy-nav tab. The thread strip already sits above the page.

6A.15 Empty states, refusals and what this page is not

The problem. A quiet page is easy to misread as “the book is fine.” These are the empty and partial states the LiveView actually paints.

What Outcomes does not do