Hierarchical learning for option implied volatility pricing

Machine learning has been a popular option implied volatility pricing approach. It brings good generalization in by avoiding building different models for options. However, it suffers from relatively low prediction accuracy besides model selection issue. In this study, we propose novel hierarchical approach to enhance machine pricing. is designed the ‘learning-hard’ problem and boosts models’ performance data on behalf of moneyness identifying optimal models. particular, proposed can be an excellent way datasets with more noise. addition, find out-of-the-money options fit better than other This pioneering study provides robust via will inspire similar studies future.