Stock Market Index Enhancement via Machine Learning

Stock market index enhancement is a popular strategy among hedge funds. The algorithm tries to adjust the weights of individual stocks of a benchmark index to boost performance of the target portfolio with respect to the original benchmark. Therefore, the key to success of this strategy is the portfolio weight adjustment procedure. In this paper, the authors extend the method in Hong et al 2021, based on machine learning and Monte Carlo simulation, to solve static constrained convex optimization problems and formulate an algorithm of stock market index enhancement. The methodology is fast, theoretically convergent under certain conditions, and able to produce superior results for high dimensional portfolio optimization problem with arbitrary convex constraints. Empirical results show that the stock market index enhancement strategy backed by the proposed algorithm can deliver stable and significant excess returns in Chinese A share markets. Therefore, the methodology proposed is of both theoretical and practical interest