As we discussed last week, momentum is among the prominent anomalies in stock markets. A momentum strategy that buys winner stocks with higher recent returns and/or shorts loser stocks with lower recent returns generates positive average profits in the short run. However, there remains questions about its validity and persistence over time and across different market states.
We approached it on our own way. We devised nine different portfolios with a combination of backward looking relative returns and the size of holdings. We looked at stocks trading on the NYSE and NASDAQ of over $1 billion in market capitalization and stocks above $5 per share. As discussed in Friday’s post, momentum strength is relative to that of the index. As such, we sorted those stocks by those with the gross excess returns compared to the S&P 500 Index. Our portfolios held the stocks with the greatest excess return. The Bloodhound dataset is a 28-year point-in-time database that is free of survivorship bias. It includes companies that traded “x” many years ago, but no longer trade today. The pricing calculations are computed daily, and thus do not have the distorting effects that many point-in-time databases have of only calculating changes and rebalances weekly or monthly.
If our ranking methodology has value, one should see improving returns as the portfolio’s concentration of holdings increases. A combination of ranks 1-10 should (in theory) outperform a portfolio of 100 equally weighted ranks if the ranking methodology has some legitimacy. The fact that our portfolios show such qualities indicates that the momentum concept appears to hold water.
Regardless of the momentun period surveyed, in each case, the 25-stock holdings outperforms the 50-stock holdings which, in turn, outperforms the 100-stock holdings. Also, as you would expect, correlation with both the S&P and the Russell 2000 increases as the porfolio becomes more diverse.
As one might expect, the Betas of the portfolios are quite high, ranging from 1.2 to 1.42x. The portfolios outperform the S&P 500 on a annual basis between two-thirds and 70% of the time. And the swings are big.
It works well, expect when it doesn’t. When the stratgies outperform the S&P, they do so significantly. However, when they underform, they do it in spades as well. Notably, the startegies were down around 60% in 2008, and to a lessor extent in 2000 and 2002. Although 20-year cumulative returns beat the S&P 500 and the Russell 2000, the volatility associated with those returns come with a higher risk. Each portfolio’s Sharpe Ratio over that same period in considerably lower than that of the index.
Note that when we shorted the evaluation time horizon from 20-years to just five, the returns lag the indicies AND the volatility remains higher. Additionally, the ranking methodology appears to be more haphazzard. Note also that standard 5-year return figures in finance no longer include 2008 in the mix.
Over a long period it appears that momentum has validity, although, like in life, the timing is considerably important. Some have taken to qualify the timing aspect. In 2004, Cooper, Gutierrez, and Hameed documented that momentum profits are higher following positive market returns than following down markets. More recently, Daniel and Moskowitz (2013) show that negative momentum profits often occur in markets where a down market is followed by an rising market. For more information, and testing of those theories, review the associated student academic paper from at the Wisconsin School of Business, University of Wisconsin-Madison which won the award for Best Doctoral Student Paper Award at the Academy of Behavioral Finance & Economics (ABFE) Meeting in 2012.
We will look at our own adjustments to the strategy later this week, as the momentum continues…