Back to Work
Trading Crypto with Reinforcement Learning
A reinforcement learning system for portfolio optimization across multiple crypto assets, robust across market regimes.
The problem
For my master's thesis, I built an AI that learned to trade cryptocurrencies and optimize portfolios across multiple assets.
The challenge was handling volatility, temporal dependencies, and risk in both bull and bear markets.
What I built / changed
- Combined Recurrent PPO with financial market domain knowledge.
- Implemented dynamic asset allocation and risk management.
- Handled temporal dependencies in volatile time series.
Result
Profitable strategy across market uptrends and downtrends.
Stack / concepts
Reinforcement LearningPyTorchTime series
Interested in discussing engineering challenges?
Get in touch