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

Background and Context

Research Question

Can machine learning methods help investors identify mutual funds that will deliver positive returns after accounting for all costs and fees?

Methodology

The study analyzes 8,767 U.S. mutual funds from 1980-2020 using three machine learning methods (elastic net, gradient boosting, random forests) to predict fund performance based on 17 fund characteristics.

Data Analysis

The researchers evaluate out-of-sample performance of portfolios constructed using the machine learning predictions, comparing them against traditional methods and passive investment strategies.

Machine Learning Methods Outperform Traditional Approaches

  • Nonlinear machine learning methods (Random Forest and Gradient Boosting) achieve significantly higher returns than traditional approaches
  • Traditional investment strategies (Equally Weighted and Asset Weighted portfolios) actually lose money after accounting for costs
  • The improvement from machine learning represents meaningful economic gains for investors

Importance of Fund Characteristics Over Time

  • The importance of different fund characteristics varies significantly over time
  • Value added and alpha t-statistic remain consistently important predictors
  • Machine learning methods can dynamically adjust to changing relationships between characteristics and performance

Fund Size vs Manager Skill Shows Misallocation

  • Top-performing funds are often smaller than expected given their manager's skill level
  • This misallocation creates opportunities for investors who can identify skilled managers
  • Machine learning helps identify funds that are "too small" relative to their manager's skill

Performance Across Market Conditions

  • Machine learning portfolios maintain positive performance across different market conditions
  • Performance is particularly strong during recessions and periods of high investor sentiment
  • The strategies show resilience and consistency across market cycles

Cumulative Performance Over Time

  • Machine learning methods demonstrate sustained outperformance over the 30-year period
  • The advantage of machine learning methods compounds over time
  • Performance advantage persists even in recent years (2019-2020)

Contribution and Implications

  • Machine learning can help investors identify mutual funds that will outperform their benchmarks after accounting for all costs
  • The success of these methods demonstrates that active management can add value when sophisticated selection tools are used
  • The findings suggest that pension fund administrators and financial advisors should integrate machine learning into their fund selection process

Data Sources

  • Method Comparison Chart: Based on Table 3 showing out-of-sample alpha of fund portfolios
  • Characteristics Importance: Based on Figure 2 showing characteristic importance over time
  • Skill vs Size Chart: Based on Figure 10 showing capital misallocation analysis
  • Market Conditions Chart: Based on Table 7 showing performance under different market conditions
  • Cumulative Performance: Based on Figure 11 showing cumulative portfolio alpha over time