# k12.energy

> Open, reproducible GB (Great Britain) power-market data and analysis: half-hourly system (imbalance) prices, market index prices, generation by fuel, Balancing Mechanism acceptances and settlements, ancillary auctions, wind curtailment, a BESS revenue index, data-quality flags and daily market summaries. Built on Elexon and NESO open data. Owned by K12 Analytics Ltd.

The site is a single-page app: fetching its HTML returns little content. The data is available directly in three ways, all read-only and without a key.

## MCP server

- [MCP server (SSE transport)](https://mcp.k12.energy/sse): 27 tools wrapping the read API, for AI assistants and agents. Each call is one bounded request; bulk files come back as links.
- [MCP documentation](https://k12.energy/api-docs#mcp)

## API

- [API documentation](https://k12.energy/api-docs): REST endpoints under https://k12.energy/v1/, JSON, date ranges half-open [from, to), list endpoints capped at 5,000 rows per page.
- [Bulk downloads](https://k12.energy/downloads): full-history CSV and Parquet per dataset, with monthly partitions and sha256 checksums.
- [Capture rates](https://k12.energy/v1/capture-rates?from=2025-11-01&to=2026-11-01&fuel=wind,solar,gas): what a MWh of wind, solar, gas, nuclear or biomass earned relative to a flat (baseload) MWh, by month, week or day, against the system price or MID. `/v1/capture-rates/cannibalisation` regresses the daily capture rate on the fuel's share of generation, with a block-bootstrap 95% interval. Method: https://k12.energy/methodology/capture-rates

## Methodology

- [Methodology](https://k12.energy/methodology): how every computed number is derived, each with a version (slug@semver).
- [Data changelog](https://k12.energy/changelog): every change to published numbers.

## Analysis

- [Labs](https://k12.energy/labs): every model and analysis on the site in one place, each with a live result, its method, API route and MCP tool: the Forecast Lab, the imbalance price rebuild, capture rates and cannibalisation, the battery revenue optimiser and settlement-based curtailment.
- [Insights](https://k12.energy/insights): monthly analysis of the GB power market by Adetunji Samuel, and daily summaries.
- [Forecast Lab](https://k12.energy/forecast-lab): leaderboard of system-price forecasts scored on the same periods at a stated issue time (09:00 London on D-1 or end of D-1): MAE, pinball loss, 80% interval coverage, skill, Diebold-Mariano tests, bootstrap intervals and error by regime. Data: /v1/forecast/backtest, /v1/forecast/predictions and the forecast-predictions download. Method: https://k12.energy/methodology/forecast-evaluation
- [Imbalance price explainer](https://k12.energy/imbalance): each period's imbalance price rebuilt from Elexon's published stack (ISPSTACK) to £0.01, with the units and fuels that set it (/v1/imbalance/explain, /v1/imbalance/price-setters, /v1/imbalance/reproduction; MCP tools get_imbalance_explanation and get_price_setters). Explained after settlement, not a forecast.

## Notes

- The market index price (MID) is a volume-weighted price of short-term exchange trades. It is not the day-ahead auction price.
- Timestamps are UTC; settlement days are Europe/London (46, 48 or 50 half-hour periods).
- Data licences and required attribution: https://k12.energy/methodology#licences
- Price model (price-model@1.0.0): a 09:00 London D-1 forecast of the GB system (imbalance) price with P10/P50/P90, built only from what was published by the issue time (Elexon NDF demand and WINDFOR wind forecast vintages, ONS gas SAP for gas days up to D-8, prices already published). It is not a day-ahead market price forecast. Model card: https://k12.energy/v1/forecast/model-card?method=price_model_linear; tomorrow's stored forecast: https://k12.energy/v1/forecast?method=price_model_linear; methodology: https://k12.energy/methodology/price-model. Gas SAP: National Gas Transmission, via ONS (OGL v3.0).