Modified IR As a Predictor of Fund Performance

40 Pages Posted: 22 Nov 2015

See all articles by Joshua Livnat

Joshua Livnat

New York University; Prudential Financial - Quantitative Management Associates

Gavin Smith

Prudential Financial - Quantitative Management Associates

Martin Tarlie

GMO

Date Written: October 1, 2015

Abstract

One of the most crucial decisions for investors and plan sponsors is the selection of funds among the thousands of available alternatives. We stress that an investor first needs to specify a target alpha, i.e., the expected fund return in excess of a benchmark, and that the target alpha determines the types of funds that are most likely to fulfill that objective. We propose a modified Information Ratio (IR), i.e., the average past alpha above both the benchmark and the target alpha, scaled by tracking error, as a screening tool for future fund performance. We find that the Modified IR is a better predictor of future fund performance than many of the other predictors in the literature, including past performance (alpha) of the fund, the fund's past tracking error, active share, the fund's intercept in regressing its return on the Fama-French factors, and that regression's R2. Consistent with intuition, we find that as the target alpha increases, "winner" funds, i.e., those in the top quintile during the past 36 months according to the Modified IR and which were also able to achieve the target alpha, have higher tracking error, fewer stock holdings on average, higher expense ratio, higher active share, higher turnover ratio, and are smaller in terms of Total Net Assets (TNA). However, we also find that the actual average annual return on "winner" funds in the year after fund selection does not increase with target alpha, and that the higher the target alpha selected by the investor the lower the likelihood of actually attaining it. Finally, for any given target alpha, past "winner" funds that were able to achieve the target alpha in the prior 36 months vary considerably in the number of positions that they have in their portfolio. However, we find that past "winner" funds with a larger number of holdings are more likely to have better future performance than those that are more concentrated and have higher active share. These results point to the advantages of performance consistency that is more likely to be achieved through diversified portfolios.

Keywords: Modified Information Ratio, Fund selection, Predicting Fund Performance

JEL Classification: G11, G14, G20

Suggested Citation

Livnat, Joshua and Smith, Gavin and Tarlie, Martin, Modified IR As a Predictor of Fund Performance (October 1, 2015). Available at SSRN: https://ssrn.com/abstract=2693607 or http://dx.doi.org/10.2139/ssrn.2693607

Joshua Livnat (Contact Author)

New York University ( email )

44 West 4th Street, Suite 10-76
Stern School of Business
New York, NY 10012-1118
United States
212-998-0022 (Phone)
212-995-4004 (Fax)

Prudential Financial - Quantitative Management Associates ( email )

2 Gateway Center
6th Fl.
Newark, NJ 07102
United States

Gavin Smith

Prudential Financial - Quantitative Management Associates ( email )

2 Gateway Center
6th Fl.
Newark, NJ 07102
United States

Martin Tarlie

GMO ( email )

United States
617-790-5072 (Phone)

Register to save articles to
your library

Register

Paper statistics

Downloads
168
Abstract Views
1,490
rank
175,590
PlumX Metrics