Projects

Statistical Arbitrage

A statistical arbitrage project on U.S. equities built around pair trading, cointegration testing, and Ornstein-Uhlenbeck modeling of the spread. The pipeline combines economic pair selection, statistical validation, dynamic hedge-ratio estimation, and signal generation through a mean-reversion state machine. It is completed with an out-of-sample backtest that includes transaction costs, risk controls, and performance analysis.

Read more

CNN Based Trading System

A CNN-based crypto trading system for BTCUSDT that turns technical indicators into 15x15 market-state images and classifies them into BUY, SELL, or HOLD signals. The full pipeline includes historical data collection, feature engineering, model training, and live inference on 15-minute Binance candles. It also includes backtesting against Buy & Hold and a real-time simulation to evaluate robustness out of sample.

Read more