Are momentum trading strategies profitable after accounting for trading costs? In their August 2007 draft paper entitled “Low-Cost Momentum Strategies”, Xiafei Li, Chris Brooks and Joelle Miffre analyze the impact of transaction costs on the profitability of momentum strategies for UK stocks. They consider all combinations of 3-month, 6-month and 12 month ranking and holding periods. Using stock price data for 3,520 UK companies and separately for the constituents of the FTSE 100 index (large capitalization stock sample) and the Alternative Investment Market (AIM – small capitalization stock sample) over the period 1986-2005, they conclude that:
- Across all strategies, before trading costs, the portfolios that are long past winners and short past losers earn an average monthly return of 1.9%, ranging from 1.6% for the 3-3 strategy to 2.2% for the 12-3 strategy. Momentum profits come mostly from the short positions, with past losers (winners) producing an average monthly return of -1.6% (0.32%). Momentum stocks, especially past losers, tend to have relatively low price, market capitalization and trading volume.
- Average total round-trip trading friction based on quoted spread is 6.7% (3.8%) for past losers (winners). These costs effectively offset the above momentum portfolio returns.
- Past losers are more expensive to trade than past winners due to the high cost of shorting stocks them (selling costs average 2.3 times higher for past losers than for past winners).
- A low-cost momentum strategy that focuses on the past winners and past losers with relatively low total transaction costs does generate abnormal returns. Momentum portfolios that are long past winners and short past losers with the 10%, 20% and 50% lowest total transaction costs generate average net annual returns of 19.1%, 15.5% and 12.6%, respectively. Among all such strategies, the 12-12-10% (12-month ranking period, 12-month holding period, among 10% of stocks with lowest transaction costs) is the most profitable with an average 27.7% net annual return.
The following chart, taken from the paper, shows the average estimated effective half-spreads for buyer-initiated (Buy) and seller-initiated (Sell) trades for three past winner portfolios, three past loser portfolios, and for the average stock in the FTSE 100 and AIM indexes. Overall, the chart shows that shorting of small capitalization past losers is costly. Specifically:
- Selling small capitalization stocks (AIM) is more costly than selling large capitalization stocks (FTSE 100), and more costly than buying any size stock.
- Selling past winners is only a little more costly than buying past winners.
- Selling past losers is much more costly than buying past losers.
Momentum strategies that ignore transaction costs tend not to outperform because of the high transaction costs of selling past losers.
In summary, successful momentum trading may depend critically on restricting consideration to stocks with the lowest total transaction costs.
