Overview
What this challenge is about.
You receive 18 months of 15-minute Nordic spot-price data, a battery dynamics model (capacity, round-trip efficiency, degradation curve), and a rule-based baseline that earns about EUR 42/MWh/day. Train a PPO policy in a custom Gym environment to choose charge/discharge actions on a continuous power setpoint. Backtest the trained policy on a held-out 3-month period, report mean daily profit, profit volatility, max drawdown, and percentage of degradation budget consumed. Write a 2-page memo for the trading desk explicitly addressing the risk of overfitting to the training period.
The Brief
What you'll do, and what you'll demonstrate.
Train and backtest a PPO bidding policy for grid-scale battery storage and quantify whether it beats the rule-based baseline net of degradation and risk.
Earning criteria — what you'll demonstrate
- Implement and tune Proximal Policy Optimization on a continuous-control problem
- Design a realistic RL environment around a physical system with degradation costs
- Backtest a learned policy with held-out time periods to detect overfitting
- Communicate RL results to a non-ML quant audience
Program Fit
Where this fits in your program.
Sharpens the same skills your degree expects you to demonstrate.
Skills
Skills you'll demonstrate.
Each one shows up on your verified credential.
Careers
Roles this prepares you for.
Real titles. Real skill bridges. Pick the one closest to your trajectory.
Applied AI Scientist
Coupling RL training with rigorous backtests and a trader-facing memo is core applied-AI-scientist work at any quant or climate-tech firm.
This challenge sharpens
- ppo
- backtesting
- risk-analysis
ML Researcher
Designing a faithful RL environment around a physical system with degradation costs is the kind of problem ML researchers tackle in industry research labs.
This challenge sharpens
- policy-gradients
- environment-design
- reinforcement-learning
Data Scientist
Walk-forward evaluation and overfitting analysis on time-series data is the data-scientist craft that transfers to any forecasting or trading role.
This challenge sharpens
- backtesting
- risk-analysis
- environment-design