A few things I've built in my own time — mostly centred around markets and ML.
Quant Dev + ML
Equity Factor Screener
Built a Python pipeline that pulls daily OHLCV data, computes momentum and
mean-reversion signals, and ranks a universe of ~200 UK & US stocks.
Used XGBoost to combine signals into a single score and backtested rankings
with a simple long-only strategy — Sharpe ~1.1 out-of-sample.
Quant Infrastructure
Python Backtesting Framework
Wrote an event-driven backtester from scratch to avoid the hidden assumptions
in off-the-shelf libraries — handles position sizing, basic slippage, and
commission.
Used it to test a handful of momentum strategies and compare results against a
buy-and-hold benchmark with rolling drawdown tracking.
Machine Learning
Return Direction Classifier
Trained an LSTM on lagged price and volume features to classify next-day
return direction on S&P 500 constituents.
Achieved ~56% accuracy on held-out data — not tradeable alone, but a useful
exercise in time-series ML, feature leakage, and proper train/test splits.
