Please rotate your device to landscape mode to view the charts.

Machine Learning and Fund Characteristics Help to Select Mutual Funds with Positive Alpha

Journal: Journal of Financial Economics

Date: 20231201

Author: DeMiguel, Victor; Gil-Bazo, Javier; Nogales, Francisco J.; Santos, Andre A. P.

Abstract:
Machine-learning methods exploit fund characteristics to select tradable long-only portfolios of mutual funds that earn significant out-of-sample annual alphas of 2.4% net of all costs. The methods unveil interactions in the relation between fund characteristics and future performance. For instance, past performance is a particularly strong predictor of future performance for more active funds. Machine learning identifies managers whose skill is not sufficiently offset by diseconomies of scale, consistent with informational frictions preventing investors from identifying the outperforming funds. Our findings demonstrate that investors can benefit from active management, but only if they have access to sophisticated prediction methods.

Link: Google Scholar

Key Findings

Machine Learning Predicts Fund Performance

Nonlinear machine learning methods select mutual fund portfolios that earn significant out-of-sample annual alphas of 2.4% net of all costs

Past Performance & Fund Activeness Interaction

Past performance is a particularly strong predictor of future performance for more active funds, revealing important interactions between fund characteristics

Capital Misallocation Effects

Machine learning identifies managers whose skill is not sufficiently offset by diseconomies of scale, consistent with informational frictions

Out-of-Sample Portfolio Performance

  • Machine learning methods (Gradient Boosting and Random Forest) significantly outperform traditional approaches
  • Nonlinear methods achieve monthly alphas of 19.7-22.4 basis points
  • Traditional passive strategies (Equally/Asset Weighted) show negative alphas

Top Fund Characteristics for Performance Prediction

  • Value added and alpha t-stat are the strongest predictors of future performance
  • Fund activeness measures (Market Beta t-stat, R²) are crucial determinants
  • Combination of past performance and activeness measures provides best predictions

Fund Size vs. Manager Skill

  • Top performing funds (D10) are smaller than expected given their skill level
  • Evidence of capital misallocation in mutual fund market
  • Informational frictions prevent optimal fund sizing

Contribution and Implications

  • First study to show investors can earn significant positive alpha using machine learning on fund characteristics
  • Demonstrates importance of nonlinear relationships and interactions in predicting fund performance
  • Provides evidence that sophisticated prediction methods can help improve mutual fund selection
  • Implications for pension plan administrators and financial advisors in fund selection

Data Sources

  • Portfolio Performance Chart: Based on Table 3 - Out-of-sample alpha of fund portfolios using FF5+MOM model
  • Characteristics Importance Chart: Based on Figure 2 - Characteristic importance analysis
  • Capital Misallocation Chart: Based on Figure 10 - Mean log size across decile portfolios