In this project, I designed an AI trading agent for electricity markets using deep reinforcement learning, trained with curriculum learning. The agent showcases strong performance in day-ahead power markets by taking in price and grid forecasts (that I developed myself using transformer models) and outputting optimal bids for the next 24 hours in a half-hourly frequency.

The agent is highly relevant for our modern energy production with an increasing share of clean, but highly volatile renewable energy source that makes operating in electricity markets very challenging. Traditional optimization methods are struggling to compete in a setting where electricity prices are fluctuating rapidly due to stochastic changes in the weather.

Ultimately, this agent could be used by power traders or grid providers as a cost-effective and highly performant trading strategy to help balance modern power grids.