Energy Modelling

Energy market forecasting and trading strategy platform built for the ENTSOE day-ahead electricity market. Includes data collection pipelines, backtesting infrastructure, a hackathon challenge framework, and a synthetic futures market model.

Terminology

Term	Meaning
Backtest	Evaluates each strategy independently using the previous day's settlement price as entry price. This is a simple, transparent baseline that students interact with directly.
Futures Market	A synthetic equilibrium model that simulates what happens when all strategies trade against each other simultaneously. Strategy weights shift based on past performance, producing a consensus "market price". This reveals whether a strategy has genuine alpha or merely rides a crowded signal.
Futures Market Simulation	The accuracy comparison tab — checks how close the futures market consensus price gets to the real day-ahead settlement.

Project Structure

src/energy_modelling/
  data_collection/   # ENTSOE & Open-Meteo data pipelines (prices, generation, load, flows, forecasts)
  backtest/          # Backtest runner, scoring, synthetic futures market engine
  futures_market/    # Shared data utilities (load_dataset, daily features, settlement computation)
  dashboard/         # Streamlit dashboard (modular: EDA, backtest, futures market, accuracy)
strategies/          # Student strategy submissions (BacktestStrategy subclasses)
tests/               # Pytest test suite
docs/                # Documentation (challenge spec, Kaggle dataset, market status)

Quick Start

Requires Python >= 3.11. Install with uv:

uv sync

Collect Data

# Requires ENTSOE_API_KEY in .env
collect-data --start 2023-01-01 --end 2024-12-31

Run Dashboard

# Single consolidated dashboard (EDA, backtest, futures market, accuracy)
streamlit run src/energy_modelling/dashboard/app.py

Run Tests

pytest

Regenerating Results

Re-run all backtests and benchmarks with a single command:

recompute-all            # all strategies × all benchmarks
recompute-all --benchmarks baseline oracle   # subset of benchmarks
recompute-all --strategies "Always Long" --verbose

Results are saved to data/results/ and picked up automatically by the dashboard.

CLI Entry Points

Command	Description
collect-data	Download ENTSOE + weather data to local parquet files
build-backtest-data	Build backtest datasets for student distribution
recompute-all	Regenerate all backtest and benchmark results

Key Features

• Data Collection: Automated pipelines for ENTSOE prices, generation, load, cross-border flows, NTC, and Open-Meteo weather forecasts.
• Backtesting: Run trading strategies against historical day-ahead prices with daily PnL tracking, Sharpe ratios, drawdown analysis, and monthly breakdowns.
• Hackathon Challenge: Framework for students to submit strategies implementing the BacktestStrategy interface, with automated scoring and leaderboard.
• Synthetic Futures Market: Equilibrium model that simulates how strategy profits change when strategies trade against each other, revealing genuine alpha vs. crowded signals.

Documentation

• Hackathon Challenge Specification
• Kaggle Dataset
• Market Implementation Status