Does a stock momentum strategy selecting only persistent winners and losers work better than a conventional strategy that includes one-month wonders? In their August 2015 paper entitled “Persistency of the Momentum Effect: The Role of Consistent Winners and Losers”, Hong-Yi Chen, Pin-Huang Chou and Chia-Hsun Hsieh examine stock momentum persistence as a condition for momentum portfolio construction. They define the momentum of a stock as persistent if it appears in the top or bottom tenth (decile) of ranking interval returns for at least two consecutive months. They determine what kinds of stocks tend to exhibit momentum persistence. They also investigate whether restricting momentum portfolios to persistent winners and losers improves performance compared to a conventional momentum portfolio. While considering several ranking intervals, they focus on six months. Using firm accounting information (lagged at least six months), stock trading data and quarterly institutional holdings for a broad sample of U.S. common stocks during 1980 through 2011, they find that:
- For a six-month ranking interval, 47% of all winner (loser) stocks exit the winner (loser) decile the next month.
- Consistent winners and losers exhibit higher subsequent momentum than inconsistent winners and losers. This difference in post-formation performance lasts more than six months.
- Momentum persistence is associated with small size, high idiosyncratic risk (volatility), low institutional ownership, high dispersion of analyst forecasts and high trading volume. These relationships suggest that persistence derives from information asymmetry and disagreement about a stock.
- A momentum strategy that selects stocks with past persistent momentum outperforms a conventional momentum strategy.
- A strategy that is long (short) the equally weighted stocks in the top (bottom) decile of past returns for two consecutive monthly six-month ranking intervals, with a skip-month between the second ranking interval and portfolio formation, generates an average gross monthly return of 1.25% over the next six months (about 15% annualized).
- A conventional momentum strategy based only on the second of the two consecutive ranking intervals generates an average gross monthly return of 1.06% over the next six months.
- Outperformance of the persistent momentum strategy survives control for market, size, book-to-market, momentum and liquidity factors.
In summary, evidence indicates that investors may be able to boost performance of a stock momentum strategy by restricting consideration to stocks that are winners for losers for at least the last two consecutive monthly ranking iterations.
Cautions regarding findings include:
- Reported returns are gross, not net. Accounting for portfolio reformation frictions and shorting costs would reduce these returns. Shorting may not be feasible for some loser stocks.
- The authors note that portfolio turnover is lower for stocks with persistent momentum than for one-month wonders. However, based on their characteristics, stocks in persistent momentum portfolios may bear higher trading frictions on average than those in conventional momentum portfolios.
- The study considers many momentum strategy variations applied to the same sample, introducing data snooping bias, such that the best-performing strategies likely overstate expectations.
- Testing variations of the momentum strategy on stock samples similar to those used in prior research adds to the aggregate snooping bias in the body of known research. To address this bias, investors should consider setting a high bar for “better” when evaluating new strategy variations.
Frequently Asked Questions
What is persistent momentum in stock investing?
Persistent momentum describes a stock whose momentum appears in the top or bottom decile (tenth) of ranking-interval returns for at least two consecutive months. Stocks meeting this condition are called persistent winners or losers, as opposed to one-month wonders that enter the extreme deciles only briefly. In the study, 47% of winner (loser) stocks in a six-month ranking interval exit the winner (loser) decile the next month.
Does a momentum strategy based on persistent winners and losers outperform conventional momentum?
A persistent momentum strategy outperformed conventional momentum in U.S. stocks over 1980-2011, generating an average gross monthly return of 1.25% versus 1.06% over the next six months (about 15% annualized for the persistent strategy). The outperformance survived controls for market, size, book-to-market, momentum and liquidity factors. Returns are gross, so trading and shorting costs would reduce them.
How is a persistent winners-and-losers momentum portfolio constructed?
A persistent momentum portfolio goes long (short) equally weighted stocks in the top (bottom) decile of past returns for two consecutive monthly six-month ranking intervals. A skip-month separates the second ranking interval from portfolio formation, and the portfolio is held for the next six months. The conventional comparison strategy uses only the second of the two ranking intervals, with data from U.S. common stocks over 1980-2011.
What kinds of stocks tend to have persistent momentum, and how do they differ from one-month wonders?
Persistent momentum stocks tend to be small, with high idiosyncratic volatility, low institutional ownership, high dispersion of analyst forecasts and high trading volume. These traits suggest persistence derives from information asymmetry and disagreement about a stock. Consistent winners and losers also exhibit higher subsequent momentum than inconsistent ones, a difference lasting more than six months, and they have lower portfolio turnover than one-month wonders.
What are the limitations of persistent momentum strategies?
Persistent momentum results are limited because reported returns are gross, and portfolio reformation frictions and shorting costs would reduce them; shorting may be infeasible for some loser stocks. Persistent momentum stocks may also bear higher trading frictions on average given their characteristics, despite lower turnover. The study tests many strategy variations on the same sample, introducing data snooping bias, so the best-performing results likely overstate expectations.