Comparing Fund Similarity Measures for Substitution and Diversification

Summary

This report compares ways to measure similarity among funds, including cosine similarity, Euclidean distance, and return correlations. It argues that the choice of measure should follow the intended meaning of “similar”: resemblance in holdings structure, similarity in holding weights, or similarity in returns. It then examines applications such as finding a substitute for a fund with restricted subscriptions, building portfolios with low market similarity, optimizing weights for diversification, estimating high-turnover funds’ industry exposures, and forecasting future return correlations.

The summarized historical comparisons report that portfolios built using return cosine similarity or return correlations had lower maximum drawdowns and higher Sharpe ratios than the cited momentum strategy. Correlation-based weighting also performed better than equal weighting in the described analysis, while return-based similarity methods helped estimate industry allocations. These findings are specific to the report’s historical data and methods; they do not establish that the relationships will persist. The report warns that market conditions and fund similarities can change, and provides no assurance of future performance.

Key ideas

• Fund similarity can refer to holdings overlap, similarity in weights, or similarity in returns, so the metric should match the use case.
• The report applies similarity measures to fund substitution, portfolio diversification, exposure estimation, and correlation forecasting.
• In its historical comparisons, return cosine similarity and return correlation produced the strongest reported Sharpe ratios among the tested methods.
• The findings depend on historical data, and changing markets or fund relationships may reduce their usefulness.

Tags

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.