Hot hands, cold hands: Does past performance predict the future?

Note: This article is abstracted from an article published by the author in the May 2006 issue of the Journal of Financial Planning.

By Dr. William G. Droms, CFA

As one key component of the mutual fund selection process, most individual investors and their advisors spend a significant amount of time studying the past performance of the funds in which they are considering investment. The “performance persistence question” simply asks: If mutual fund performance does not persist, can there be any possible justification for including past performance as a key criterion for selecting mutual funds?

A review of 10 major studies published during the last 15 years on the topic of performance persistence provides a useful set of lessons learned from these studies that can be used by practitioners to structure mutual fund investment portfolios for clients.

Many of the tests of mutual fund performance persistence test for the persistence of risk-adjusted performance, most often via tests of persistence of portfolio “alpha.” Alpha for a mutual fund is a measure of the return on the fund in excess of the return predicted by capital asset pricing theory as a function of the systematic risk of the fund and the return on the market. Statistically, alpha is computed by fitting a regression line regressing the return on the fund against the return on an appropriate market index, such as the stock market in general (the S&P 500 Index, for example) or a style-based benchmark (such as an index of value stocks or an index of growth stocks). The return measure normally used is the “excess return” of the fund compared to the “excess return” on the index, where excess return is calculated as the return on the fund or index minus the risk-free rate of return. Alpha is the intercept of this regression line, and thus measures the excess return earned by the fund when the excess return on the index is zero.

Funds with positive alphas that are significantly greater than zero are judged to be outperforming the market. Tests for persistence of alpha thus test for persistence of performance above or below that predicted by the riskiness of the fund.

Tests of performance persistence commonly employ a successive “winner-winner, winner-loser” methodology whereby funds are rank-ordered by one-year total returns. The 50 percent of the funds with the highest returns are labeled “winners” and the 50 percent of the funds with the lowest returns are labeled “losers. Funds that ceased operations in the subsequent are labeled as “gone.” Statistical tests are used to test whether the number of funds with persistent performance (i.e., the number of winner-winner and loser-loser pairs) is significantly greater than what would be predicted by chance.

A number of conclusions, or at least inferences, can be drawn from the review of performance persistence studies:

Past performance counts: The studies reviewed present generally consistent evidence that past performance persists, at least in the short run. The studies find with remarkable consistency that superior performance in one year is predictive of superior performance in the following year. Furthermore, studies that test investment strategies of investing in the previous year’s superior performing funds find that the superior performing funds outperform the funds in aggregate in the following year and that they substantially outperform portfolio of inferior performing funds. In short, hot hands persist, albeit for short time periods.
Poor past performance counts more than good past performance: Persistence of the “cold hand” phenomenon is the strongest and most consistent conclusion found in all of the major studies: Poor past performance is a strong predictor of future poor performance.
Within-category performance is more persistent than performance relative to overall market: Within-category performance is more persistent than superior performance relative to overall market. Mutual fund performance should be evaluated relative to a style-based benchmark rather than relative to the market in general when evaluating performance or performance persistence. It is quite likely, for example, that a superior performing large-cap value fund would under-perform the stock market in general in a year in which growth stocks outperform value stocks, even if the value fund consistently outperforms the average value fund.
Short-term performance persistence is much stronger than long-term persistence: The evidence is less clear on long-term performance as compared to short-term performance. The studies differ on whether performance persists in the long term, although the balance of the evidence leans in the direction of much weaker persistence or no persistence over longer time periods.
International equity funds show very strong short-term performance persistence but no longer-term persistence: The one study specifically testing persistence of international equity funds finds highly significant performance persistence for one-year holding periods, but no evidence of persistence for holding periods of two, three or four years.
Morningstar within-category ratings provide useful information for selecting mutual funds based on past performance: Morningstar star ratings for mutual funds are based solely on past performance. One study produced strong evidence that low Morningstar ratings are reliable predictors of future poor performance. Four- and five-star funds were outperformed by median funds in subsequent holding periods.

The most important practical implication from performance persistence studies is that the historical performance of mutual funds contains useful information about future performance, at least for one-year holding periods. This is not to say that historical performance should be the sole or even primary criterion for selecting mutual funds, but it does recognize that past performance provides significant information about likely future performance. Historical performance can provide a useful screen to identify likely candidates for future investment as well as a list of likely candidates for sale.

The results also demonstrate that performance should be judged against style-based market benchmark indexes. Combining these results with the Morningstar results tells us that the Morningstar star system, based on style-based comparisons, is a useful system to develop likely candidates for investment or avoidance. And, in fact, the results indicate that any screening system based on fund performance relative to an appropriate style-based index should provide useful lists of candidates for investment across the various investment styles.

Finally, the results demonstrate that sub-par performance in one year is an important red flag that, at a minimum, should alert mutual fund investors to the possibility of selling an underperforming fund. However, since even those funds with excellent long-term performance almost always experience one or more years of sub-par performance, one year of below-average performance should not be the sole or even the most important criterion for selling a fund.

In summary, past performance, especially poor past performance, provides useful information about expected future performance.

Dr. William G. Droms, CFA is the John J. Powers, Jr., professor of finance in the McDonough School of Business at Georgetown University and co-chair of Droms Strauss Advisors, Inc., a registered investment advisor. He has published numerous articles in professional journals, including the Financial Analysts Journal, Journal of Accountancy, Journal of Financial Planning and CPA Journal.

This content has not yet been Rated.

To Rate content, please Login.