Comparing Mutual Fund Styles with Weighted Portfolios and Factor Alphas

Summary

The document proposes a way to compare mutual funds grouped by holdings-based investment style. For each style, construct both an equally weighted portfolio and a value-weighted portfolio from the funds’ quarterly returns. These alternative aggregations show whether findings depend on treating each fund equally or giving greater influence to larger funds.

Next, regress each style portfolio’s returns on a factor model, with a four-factor model and additional factors relevant to mutual fund performance suggested as a starting point. Assess whether each portfolio’s estimated alpha differs from zero, then test whether the alphas differ across styles to address relative performance. The response gives a compact research outline, not a complete empirical design: it does not specify factor definitions, standard error treatment, survivorship or fee adjustments, or tests for changing exposures. A return plot may be useful for description, but the proposed inference relies on factor-adjusted estimates and explicit alpha comparisons.

Key ideas

• Aggregate each style’s fund returns into both equally weighted and value-weighted portfolios.
• Use a factor model to estimate risk-adjusted returns for each style portfolio.
• Test whether each portfolio’s alpha differs from zero.
• Test differences between style portfolio alphas to evaluate relative performance.
• The outline leaves data-quality and statistical-inference choices to the researcher.

Tags

# Mutual fund performance over time


# Mutual fund performance over time












I am a little bit stuck with my dissertation thesis, so help will be greatly appreciated. I am trying to analyze the performance of different mutual funds, which I have classified according to investment style using a holdings-based analysis. My sample period of reference ranges from 2007 to 2018, with quarterly observations. I have also retrieved the quarterly returns for all these mutual funds, over the sample period.

Now I would like to aggregate all the funds which have a similar investment strategy, and compare the performance of these different groups over the period, eventually demonstrating whether the difference in performance is statistically significant or not.

Do you have any suggestion on how should I aggregate the returns? And as for the comparison of performance, should I simply plot the returns and run a regression and then look at the coefficients?

Thanks in advance

## Answer by phdstudent (score 1)

https://quant.stackexchange.com/a/55241

My suggestion is the following:

- Build a value-weighted and equally weighted portfolio of mutual fund returns based on their style.

- Regress the returns on all portfolios against a factor model (ideally 4-Factor model + factors known to explain mutual fund performance)

- Se if $\alpha$'s are different from zero.

- Test for differences in $\alpha$'s (performance)

Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)

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