Forex Forecasting and Option Hedging

Completed during my internship at BIAT - Banque Internationale Arabe de Tunisie (Tunis, Tunisia). The project combined econometrics and machine learning to forecast currency markets and develop hedging strategies.

Data Analysis

Processed and analyzed currency rate datasets to identify market patterns

Forecasting Models

Built econometric and machine learning models to predict forex rates and market volatility

Hedging Strategies

Designed and evaluated an option hedging strategy, comparing its performance against Delta Hedging

Key contributions

• Model benchmarking: Benchmarked 10+ time series models (ARIMA, LSTM, SVR, XGBoost, ensemble methods) for currency rate and volatility forecasting across 2 currency pairs—LSTM achieved RMSE 0.0052 for currency prediction through hyperparameter optimization of 480+ configurations, outperforming econometric benchmarks.
• ML-based hedging: Developed ML-based options hedging strategy combining predicted rates and volatility to forecast delta—backtested across multiple scenarios (maturities, time horizons, strike prices), reducing hedging costs by ~3.5% compared to traditional delta hedging.