How it works: The machine-learning model uses forecast grid demand, generation, capacity, and weather to predict future prices based on training with historical prices.
When generation exceeds demand (usually due to high wind), prices tend to drop. When demand signficantly exceeds generation by renewables sources, generation by gas must be increased, and prices tend to increase.
Day-ahead prices are released around 10:00. Actual prices from Octopus are released around 16:00.
Forecasts are run every 3 hours, starting at 01:05. All chart times are London local time. API documentation