Quant / Built
Statistical Arbitrage Research Platform
Investigating pairs, changing hedge ratios, and the assumptions behind a backtest.
- My contribution
- Implemented cointegration tests, dynamic hedge ratios, and backtest risk analysis.
Python / Streamlit / Kalman filter
The problem
Pairs research requires inspection of both the price relationship and trading assumptions.
Implementation
The platform combines Engle Granger and Johansen cointegration, Kalman filter hedge ratios, and Ornstein Uhlenbeck estimation. Streamlit exposes backtesting and risk analysis.
Evaluation & limitations
Backtests depend on transaction costs, lookahead bias, and market conditions. Reproducible performance results are not yet available.