Atlas Support
ResearchFinanceWritten and edited by:

Validating the Value-Add of Machine Learning in Multi-Strategy Portfolio Management

Atlas Support’s Economic Research team introduces research testing whether machine learning adds value over equal weighting and risk parity when allocating across investment strategies.

Three allocation methods across the same 10 strategies

The study uses daily returns for 10 strategies from February 2022 through November 2025 and compares three portfolios with monthly allocations based on equal weighting, risk parity, and machine learning.

Risk-adjusted performance in the sample

The machine learning allocation recorded the highest Sharpe ratio of the three and an information ratio of about 0.94 relative to risk parity. The comparison is specific to the sample and period. The research page identifies a comparison with total-return TOPIX and a factor-based analysis as follow-up work.