Enhancing Passive Index Investing: Augmenting Volatility Harvesting with Machine Learning Models for Higher Returns

Active fund management has traditionally dominated investment strategies but has increasingly faced scrutiny due to its inability to consistently outperform benchmark indices, prompting a shift toward passive index investing. While passive strategies simplify investment by mirroring market returns, they inherently cap the potential for higher gains and expose investors to market volatility. This study explores how dynamic portfolio rebalancing, combined with predictive insights from CNN-GRU hybrid and transformer-based machine learning models, can enhance returns while preserving the benefits of index-based investing. By employing volatility harvesting and market timing strategies at various time scales, the proposed approach augments portfolio weights dynamically (by a small amount of 2.5%), favoring equity indices during bullish predictions and reducing exposure during bearish forecasts. Through this methodology, we demonstrate improved risk-return profiles, highlighting the untapped potential of augmented volatility harvesting as a cost-effective, AI-driven investment strategy.