Abstract
In recent years, many studies have explored artificial intelligence (AI) in quantitative trading and financial prediction. Among financial products, options are highlighted for their risk management capabilities and defined trading cycles, yet they pose challenges due to their complexity. This paper proposes HAPPY (Heatmap Analysis via PON/POD Yield), an options trading system designed to predict expected winning rates (EW) by integrating actual profits, losses, and risk factors, thus enhancing traditional winning rate metrics. HAPPY employs heatmap analysis to address the risk of overfitting by smoothing isolated low EW values and incorporates machine learning (ML) models like random forest, extreme gradient boosting (XGBoost), and light gradient boosted machine (LGBM) for improved prediction accuracy. Employing TAIEX weekly options, the study evaluates EW and backtests trading performance, comparing empirical statistics and ML models. Findings indicate that ML models excel in accuracy and precision, though empirical statistics perform better in backtesting, especially as options near expiration. This research offers a robust options trading system that can be applied to other options markets or predictive models.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call