Blog - Investing Notes

March 25, 2009 - Update: Beat the Market with Hot-Anomaly Switching?

Can investors beat the market by iteratively finding and exploiting the current hot anomaly? In the February 2009 version of his paper entitled "Real-Time Profitability of Published Anomalies: An Out-of-Sample Test", Zhijian Huang investigates whether a trader can realize excess returns by repeatedly picking the anomaly with the best return during a rolling historical window from an expanding universe of anomalies as published, with a specific objective of suppressing data snooping bias. The universe includes anomalies that: (1) have been published in at least one of three top-ranked finance journals; (2) relate to the calendar or to cross-sectional predictability; and, (3) can be re-evaluated annually. Using monthly return data associated with 11 anomalies published during 1977-1991 (Monday effect, January effect and cross-sectional effects related to size, book-to-market ratio, momentum, earnings/price ratio, cash flow/price ratio, dividend yield, debt/equity ratio, sales growth and trading volume/turnover) as available from 1926 through 2006, he concludes that:

• A trader can outperform the market by by an average 4.43% to 13.23% annually, based on 1% round-trip trading frictions, by iteratively switching each year to the best past performer among published anomalies. (See the chart below.)
• Excess returns are highest for past performance (training) periods of two to five years.
• While the January effect and momentum effect are drivers, similar results accrue from consideration of only a subgroup of anomalies. Increasing the number of anomalies considered steadily boosts average realized excess returns.
• Outperformance of benchmark is robust to different levels of transaction costs and to consideration of subgroups of anomalies rather than the entire group of 11.
• Returns for nine of 11 anomalies decline after publication (see the table below). However, anomalies still generally retain value within the context of the hot anomaly selection strategy.

The following chart, taken from the paper, compares the cumulative wealth from $1 initial investments in 1977 based on the best performing anomaly over past one, two, five, and ten years, retested/reselected annually. It shows that a five-year past performance (training) period generates the greatest excess returns. All results include 1% round-trip trading frictions.

The following table, excerpted from the paper, compares the annual returns of the 11 anomalies before and after publication: Monday effect, January effect (JanEffect) and cross-sectional effects related to size, book-to-market ratio (B/M), momentum, earnings/price ratio (E/P), cash flow/price ratio (CF/P), dividend yield (Div/P), debt/equity ratio (D/E), sales growth (GS) and trading volume/turnover. Excess Before (Excess After) is the average annual return in excess of the the benchmark return before (after) the year of publication. Relative performance declines post-publication for nine of 11 anomalies. Data for the Volume anomaly is limited.

In summary, a trader who periodically switches to the hottest known anomaly based on a rolling window of past performance may be able to beat the market. Anomalies appear to have their own kind of momentum.

Perhaps the very best trading strategies are those that rotate (but not too often, based on trading costs) to the hottest anomalies in the hottest asset classes. See our blog entries of 1/7/08, 7/27/07 and 2/27/07 for strategies that use one or two anomalies to rotate through asset classes. See also our blog entry of 2/7/08 for evidence that inflexible strategies work only for limited periods.

For other big-picture research, see Blog Synthesis: Big Ideas for Investing/Trading.