Documentation — Grid Poet

1. DA Price Forecast

Grid Poet forecasts day-ahead electricity prices for the DE-LU bidding zone (Germany + Luxembourg). The forecast is generated daily before the EPEX SPOT day-ahead auction closes at 12:00 CET, predicting 24 hourly prices for the next delivery day (D+1).

The model outputs three quantiles: P10 (10th percentile, lower bound), P50 (median, point forecast), and P90 (90th percentile, upper bound), forming an 80% prediction interval.

1.1 Pipeline Flow

The daily forecast pipeline runs in five stages:

Daily Forecast Pipeline ┌─────────────────┐ ┌─────────────────┐ ┌─────────────────┐ │ ENTSO-E │ │ ENTSO-E │ │ Open-Meteo │ │ DA prices │ │ Generation mix │ │ Weather fcst │ │ (D-10 → D-1) │ │ Consumption │ │ 6 locations │ │ hourly EUR/MWh │ │ Residual load │ │ (D-10 → D+1) │ └────────┬────────┘ └────────┬────────┘ └────────┬────────┘ │ │ │ └────────────────┬───────┴────────────────────────┘ ▼ ┌────────────────────────┐ │ 1. FETCH & MERGE │ │ Align to UTC hourly │ │ Deduplicate sources │ └───────────┬────────────┘ ▼ ┌────────────────────────┐ │ 2. FEATURE ENGINEERING │ │ Weather aggregation │ │ Calendar features │ │ Price/gen lags (D-1) │ │ Derived (wind³, solar) │ └───────────┬────────────┘ ▼ ┌────────────────────────┐ │ 3. PREDICT │ │ LightGBM × 3 quantile │ │ → P10, P50, P90 │ │ for 24 delivery hours │ └───────────┬────────────┘ ▼ ┌────────────────────────┐ │ 4. STORE │ │ Save forecast + meta │ └────────────────────────┘ Actuals Scoring (after auction) ┌────────────────────────┐ │ 5. SCORE │ │ MAE, RMSE, bias │ │ 80% coverage check │ └────────────────────────┘

The forecast is generated before the EPEX SPOT day-ahead auction closes at 12:00 CET. All input data is available by forecast time: historical prices up to D-1, ENTSO-E generation up to D-1, TSO wind/solar forecasts for D+1, and weather forecasts for D+1.

1.2 Data Sources

Source	Data	Resolution
ENTSO-E Transparency Platform	DA prices, generation by source, load, cross-border flows, wind/solar forecasts (DE-LU + 13 zones)	15-min / hourly
Energy-Charts (Fraunhofer ISE)	Generation mix (fallback source, publishes faster than ENTSO-E for some types)	Hourly
Open-Meteo	Wind speed (100m), solar radiation, temperature, cloud cover	Hourly

Weather Locations

Weather data is fetched from 6 representative locations across Germany, weighted by installed renewable capacity:

Location	Coordinates	Represents
North Sea	54.5°N, 7.0°E	Offshore wind cluster
North	53.5°N, 10.0°E	Onshore wind (Schleswig-Holstein, Niedersachsen)
Central West	51.5°N, 7.5°E	NRW industrial demand, onshore wind
Central East	51.3°N, 12.4°E	Sachsen-Anhalt / Brandenburg wind+solar
South West	48.8°N, 9.2°E	Baden-Württemberg solar
South East	48.1°N, 11.6°E	Bavaria solar

Wind speed is weighted by installed wind capacity (offshore/onshore splits). Solar radiation is weighted by installed PV capacity (concentrated in southern Germany). Temperature uses equal weighting.

1.3 Features (~37 total)

Weather (D+1 forecast)

Capacity-weighted wind speed at 100m hub height
Capacity-weighted direct solar radiation
Average temperature and cloud cover across locations
North–south wind gradient (congestion proxy)
North–south radiation gradient

Price Lags (strict no-leakage: D-1 and earlier)

Same hour yesterday (D-1) and one week ago (D-7)
Yesterday's daily mean, min, max
7-day rolling mean and standard deviation

Generation Lags (D-1)

Gas, coal, and renewable generation shares
Consumption and residual load
Gas-to-coal ratio (implicit fuel economics proxy)

Calendar

Hour, day of week, month
Weekend, public holiday (all 16 German federal states), bridge day
Cyclical encodings (hour sin/cos, month sin/cos)

Derived

Wind power proxy — wind speed cubed (cubic power law, capped at 25 m/s cut-out)
Solar capacity factor proxy — radiation adjusted for temperature derating
Residual load × peak hour interaction

No-leakage guarantee: All lag features use only data from D-1 or earlier. Weather features use D+1 forecasts (available on the morning of D). No feature uses actual data from the delivery day.

1.4 Model

LightGBM gradient-boosted decision trees with quantile regression objective. Three separate models predict P10, P50 (median), and P90, sharing the same feature set.

A single model covers all 24 hours with hour as a feature, rather than 24 separate hour-specific models. This gives the model more training data per hour and lets it learn cross-hour patterns.

Training approach

Historical data from January 2024 to present. Walk-forward evaluation: train on all data except the last 30 days, test on the final 30 days. Hyperparameters are tuned to minimize MAE on the held-out test set.

1.5 Evaluation

Metrics

Metric	Description
MAE	Mean absolute error — average distance between forecast and actual price
RMSE	Root mean squared error — penalises large errors more heavily
Mean bias	Average (forecast − actual). Positive = systematically overestimating
80% coverage	Fraction of actuals falling within the P10–P90 interval. Target: 80%
Skill score	1 − (MAE / baseline_MAE). Positive = better than persistence (same-hour D-1)

Baseline

The persistence baseline predicts tomorrow's price = today's price for the same hour. This is the simplest reasonable forecast and the benchmark to beat. A skill score of 0.40 means the model reduces MAE by 40% compared to persistence.

Performance metrics and detailed error analysis are available on the forecast page.

1.6 Timestamp Conventions

Power market timestamps are interval-aligned: 14:00 means the interval from 14:00 to 15:00 (left-aligned). All internal timestamps use naive UTC.

DST transitions are handled explicitly: a CET delivery date may have 23, 24, or 25 hours. Display conversion to Berlin time is applied in the frontend.

2. Grid Poet Image Pipeline

The image pipeline runs hourly and transforms live grid data into AI-generated artwork:

Fetch — current generation mix, consumption, weather from ENTSO-E + Open-Meteo
Analyse — Claude AI writes an analysis, poem, and image prompt based on the grid state
Generate — FLUX.2 Pro creates a painting in the 19th-century German Romantic landscape tradition
Publish — image + metadata saved, served on the main page

Each image captures a moment in Germany's energy transition — the interplay of wind, solar, fossil fuels, and weather — through the lens of landscape painting.

3. Captured Prices & Cannibalization

The Analytics page shows captured prices and generation-vs-price scatter plots using hourly market data.

Source data

Data	Source	Unit
DA price	ENTSO-E	EUR/MWh
Wind onshore	ENTSO-E	GW
Wind offshore	ENTSO-E	GW
Solar	ENTSO-E	GW
Biomass	ENTSO-E	GW
Hydro	ENTSO-E	GW
Natural gas	ENTSO-E	GW
Hard coal	ENTSO-E	GW
Brown coal (lignite)	ENTSO-E	GW
Consumption (total load)	ENTSO-E	GW

Derived metrics

Metric	Formula
Captured price (per technology)	Σ(price_h × gen_h) / Σ(gen_h) — volume-weighted average DA price
Baseload price	Σ(price_h) / N — simple arithmetic mean, the reference line
Net import	max(0, consumption − total generation) — positive during import hours

Why captured prices matter: When wind and solar generate the most, they depress market prices (the merit order effect). The captured price shows the actual revenue per MWh each technology earns — typically below baseload for renewables. This "cannibalization" is a key metric for renewable economics.

4. European Markets

The Markets page compares day-ahead electricity prices, generation mixes, and cross-border flows across 14 European bidding zones: Germany (DE-LU), France (FR), Netherlands (NL), Austria (AT), Poland (PL), Denmark West (DK1), Belgium (BE), Czech Republic (CZ), Spain (ES), Portugal (PT), Norway South (NO1), Sweden Mid (SE3), Finland (FI), and Great Britain (GB).

Data sources

Day-ahead prices for continental zones are sourced from the ENTSO-E Transparency Platform (authoritative TSO data). Generation mix data is sourced from Energy-Charts (Fraunhofer ISE, CC BY 4.0). GB uses Elexon BMRS (settlement prices, not DA auction).

Timestamps

All timestamps refer to the delivery interval — the time during which power was physically delivered. Timestamps are left-aligned: "14:00" means 14:00–15:00.

Cross-border flows

Cross-border data shows scheduled commercial exchanges for Germany. Positive values = export from DE, negative = import to DE. Source: ENTSO-E Transparency Platform (values in GW).

5. Data Attribution

ENTSO-E — Transparency Platform: DA prices, generation, load, cross-border flows (14 zones)
Energy-Charts — Fraunhofer ISE: generation mix fallback (CC BY 4.0)
Elexon — BMRS: GB settlement prices
Open-Meteo — open-source weather API (ECMWF, DWD, and other NWP models)
FLUX.2 Pro — Black Forest Labs: image generation
Claude — Anthropic: analysis, poetry, and image prompts