Programmes

Learning Programmes

Structured self-study in gas and power. Each market ships as a workbook that recalculates and reconciles, plus a deck that cites its sources. Every figure carries a verification code, and nothing is estimated into a gap.

European Gas and Power — Twelve markets on one template, built so they are directly comparable. Market architecture, the regulatory perimeter, the carbon overlay, and a workbook per market that recalculates and reconciles rather than asserting.

Last reviewed 30 Jul 2026

60 hrs

Study budget across four weeks

Market packs, each a workbook and a deck

Power Query definitions, live against primary sources

Open data gaps — every one named, none estimated

Batch structure

Batch	Scope	What it covers	Status
0	Foundation and reference library	Target model, day-ahead and intraday coupling, balancing, the gas entry-exit regime, the REMIT / EMIR / MiFID II perimeter, EU ETS and CBAM, Fit-for-55, storage and LNG.	Complete
1	Netherlands · Germany · France · Belgium	Corrected and rebuilt after a cross-batch reconciliation found all four markets failed a components-versus-total check — Germany by 17.51%.	Complete, rebuilt
2	United Kingdom · Italy · Spain · Nordics	GB carbon price support with a dated switch at April 2028; Nordic hydro reservoir balance and the twelve-zone EPAD table.	Complete
3	Consolidation	Eleven markets merged onto one basis. The cross-batch harness is complete and passing; the consolidated workbook and deck are outstanding.	Part complete
4	Quant layer	Six working models calibrated on the market structure from the earlier batches.	Planned

What the build actually found

The defect that justified the whole verification standard — Batch 1 shipped with no components-versus-total reconciliation. When one was finally applied during consolidation, all four markets failed. Germany was out by 90,028 GWh — coal sub-classes had been listed alongside their parent class and double counted, while three other fuels were omitted entirely.Root cause — two Eurostat datasets treat pumped hydro oppositely, and one mental model had been applied to both.

A verification step that cannot fail is worse than none — The harness was passing vacuously. LibreOffice was writing its output over its own input, so "zero errors" was being measured on a workbook that had never been recalculated.It now asserts every formula carries a cached result before asserting anything about the results.

North American Gas and Power — Seventeen packs across seven US ISOs, five gas hubs, Canada and Latin America. Built to the same standard as the Europe programme and deliberately broken from it wherever the market is structurally different — which is most places.

Last reviewed 1 Aug 2026

80 hrs

Study budget across six weeks

Market packs planned

Registrations required — the build runs keyless end to end

Failure modes pre-registered, each with the control that catches it

Batch structure

Batch	Scope	What it covers	Status
0	Foundation — power and gas	Two decks, thirty slides. The FERC / NERC / state stack, nodal LMP decomposed, congestion rights, capacity versus energy-only, the load shock, four carbon regimes. Plus Henry Hub as a physical point, Order 636, basis as the traded object, and the unit conventions that break European formulas.	Complete and verified
1	PJM · ERCOT · MISO	The three archetypes, not the three biggest — a nodal market with a capacity construct, an energy-only market of scale, and a market defined by its seams.	Next
2	CAISO · NYISO · ISO-NE · SPP	Variations on the Batch 1 archetypes, which is why they compress into the same hours.	Planned
3	Five gas hubs	Henry Hub / Gulf, Permian–Waha, AECO / WCSB, Northeast, West. Gas gets its own batch because the tradable object is a differential between two physical points, which belongs to no single power market.	Planned
4–5	Canada and Latin America	Alberta and Ontario; Mexico, Brazil, Chile and Colombia. Brazil reuses the Nordics template — the one place a European pack transfers unmodified.	Planned
6–7	Environmental markets, consolidation, quant layer	RGGI, the WCI link, Washington's CCA, RECs and SRECs, and IRA credits read as a negative carbon price.	Planned

Download

Foundation 0A — North American Power Markets

Nodal pricing, resource adequacy and a fragmented carbon overlay. Fifteen slides with speaker notes.

0.7 MB · PPTX

Foundation 0B — North American Gas

One benchmark, a continent of basis, and the export pull that now sets the margin. Fifteen slides with speaker notes.

0.7 MB · PPTX

What the build actually found

The 10% error waiting for any European practitioner — European gas works on a lower-heating-value basis. North American gas is priced, sold and custody-metered on higher heating value, and US heat rates are quoted the same way. A European spark-spread formula applied unchanged inserts a correction factor that does not belong there.Measured — adding the European 0.9025 factor moved the ERCOT test case from $17.80 to $19.92/MWh, an 11.9% overstatement.

Same physics, different basis — The Europe programme's 0.20196 tCO2/MWh on an LHV basis converts to 53.42 kg/MMBtu HHV, against EPA's published 53.06. They agree to 0.67%.The workbook carries that bridge as live formulas and fails the build if the agreement exceeds 1%.

Gate 0 cleared without a single registration — Every blocked data source was probed for a keyless equivalent. EIA's bulk archive replaced the keyed API entirely; MISO's failed real-time route was replaced by its market-report file server; CAISO only ever needed an unzip step.EIA publishes its hourly archive as "older than 7 days" — history, not the last week. That caveat is carried in three places so it cannot be lost.

The standard both programmes are held to

Reconcile every dataset before building anything from it. Recalculate every workbook and assert no cell errors — but assert first that every formula carries a cached result, because a verification step that cannot fail is worse than none. Render every slide and look at it. Check the arithmetic by hand. Name every gap and estimate into none of them.