Stock Momentum Exploiting All Price Data in a Lookback Interval

Does use of price data other than the first and last within a lookback interval improve performance of a stock momentum strategy? In their November 2022 paper entitled “Momentum Without Crashes”, Soros Chitsiripanich, Marc Paolella, Pawel Polak and Patrick Walker construct a momentum strategy that ranks stocks based on a weighting scheme using prices throughout the lookback interval, in effect combining reversal and momentum patterns in returns. Specifically, they apply fractional differencing to stock price series differencing parameter d ranging from 0 to 1. When d is 1 (0), the result is a conventional momentum (reversal) strategy. A value of d between 0 and 1 combines momentum and reversal signals. Each week they sort stocks into fifths, or quintiles, by ascending expected returns based on a specific value of d and a lookback interval of 250 calendar days (one year). They then construct a value-weighted or an equal-weighted portfolio that is long (short) the quintile of stocks with the highest (lowest) expected returns. To avoid any day-of-the-week effects, they construct  such portfolios each weekday and average returns across five weekly-reformed portfolios. They consider a sample of all U.S.-listed common stocks and a subsample that selects only stocks that comprise the top 90% of of market capitalization that week (excluding small stocks). For robustness, they consider smaller/shorter samples from six other countries. Using daily prices for the specified stock samples as available during January 1972 through December 2020, they find that: (more…)