Maximizing Portfolio Predictability with Machine Learning
Using machine learning to optimize stock portfolios for predictability, outperforming benchmarks with various models and strategies.
What it examines
This paper revisits the maximally predictable portfolio (MPP) problem, using machine learning models to forecast stock returns and compute optimal portfolio weights under realistic constraints. The study aims to enhance portfolio predictability and performance using advanced algorithms and machine learning techniques.
What it concludes
The study concludes that machine learning-based MPPs offer significant improvements in portfolio predictability and performance. Potential applications include enhanced investment strategies and better risk management. Future research could explore additional machine learning models and further refine the optimization algorithms.
Evidence objects
The study concludes that machine learning-based MPPs offer significant improvements in portfolio predictability and performance. Potential applications include enhanced investment strategies and better risk management. Future research could explore additional machine learning models and further refine the optimization algorithms.
key_findings bullet 1 · key_findings · validation V0
Raw abstract and provenance
Abstract: We construct the maximally predictable portfolio (MPP) of stocks using machine learning. Solving for the optimal constrained weights in the multi-asset MPP gives portfolios with a high monthly coefficient of determination, given the sample covariance matrix of predicted return errors from a machine learning model. Various models for the covariance matrix are tested. The MPPs of S&P 500 index const… ▽ More We construct the maximally predictable portfolio (MPP) of stocks using machine learning. Solving for the optimal constrained weights in the multi-asset MPP gives portfolios with a high monthly coefficient of determination, given the sample covariance matrix of predicted return errors from a machine learning model. Various models for the covariance matrix are tested. The MPPs of S&P 500 index constituents with estimated returns from Elastic Net, Random Forest, and Support Vector Regression models can outperform or underperform the index depending on the time period. Portfolios that take advantage of the high predictability of the MPP's returns and employ a Kelly criterion style strategy consistently outperform the benchmark. △ Less
Source row: 1305 · abstract type: unknown