Asset Allocation via Machine Learning and Applications to Equity Portfolio Management

TL;DR

This paper introduces a machine learning-based approach for portfolio optimization that avoids traditional covariance matrix computations, enabling effective asset allocation with significant excess returns demonstrated in U.S. and China markets.

Contribution

The paper presents a novel machine learning method for asset allocation that simplifies optimization and improves returns without relying on covariance matrix estimation.

Findings

Achieved significant excess returns over benchmarks.

Eliminated the need for covariance matrix inversion.

Demonstrated effectiveness in U.S. and China equity markets.

Abstract

In this paper, we document a novel machine learning based bottom-up approach for static and dynamic portfolio optimization on, potentially, a large number of assets. The methodology applies to general constrained optimization problems and overcomes many major difficulties arising in current optimization schemes. Taking mean-variance optimization as an example, we no longer need to compute the covariance matrix and its inverse, therefore the method is immune from the estimation error on this quantity. Moreover, no explicit calls of optimization routines are needed. Applications to equity portfolio management in U.S. and China equity markets are studied and we document significant excess returns to the selected benchmarks.