In May 2015, Huseyin Gulen and Ralitsa Petkova published "Absolute Strength: Exploring Momentum in Stock Returns"  (SSRN).

In the paper they outline their new concept of absolute strength momentum.

Momentum, in its traditional form, was a relative strength concept.  Momentum took the cross-section of returns across securities, bough the "winners" and sold the "losers."  In this traditional measure, there is no consideration for whether the securities are increasing or decreasing in value: only a relative comparison between security performance.  So losers may actually be appreciating in value; winners are just appreciating faster.

Gulen and Petkova proposition is to split securities based on their absolute return levels: winners are those that have had a statistically significant positive return and losers are those that have had a statistically significant negative return.

This is akin to traditional time-series momentum, but instead of using zero, or a risk-free rate, as the threshold, the authors establish a higher hurdle.

To be classified as an absolute strength winner (loser), a stock must have a recent cumulative return in the top (bottom) 10% of the historical cumulative return distribution.

The underlying philosophy behind this strategy is that strong absolute returns, in either direction, play an important role in both rational and irrational investor behavior.  Gulen and Petkova provide many potential reasons including: tax-loss selling, capital gains overhang, loss aversion, anchoring, and mental accounting.

But, perhaps the strongest potential reason is the disposition effect, where investors become more risk averse following losses and less risk averse following gains.

Corey here: I think this may also be related to the rational inattention hypothesis for momentum.  For example, Chen and Yu (2014) found that portfolios constructed from stocks “more likely to grab attention” based on visual patterns induces investor over-reaction. They provide evidence that momentum continuation is is induced by visually-based psychological biases.

A unique aspect of this study is that, unlike relative strength momentum, an absolute strength momentum portfolio may not always have stocks designated as winners or losers.  There may be periods where there are not enough securities (defined by the authors to be 30 holdings) to fill one leg of the trade, creating an unhedged portfolio.  In this scenario, the authors invest in 1-month T-bills.

The authors argue that in such a scenario, it is likely the market has moved too far in one direction and an relative momentum crash is highly probable.  This occurs because relative strength sorting is highly tied to overall market momentum.  For example, in market crashes, it is likely that the relative winners (losers) are low (high) beta securities; in a subsequent market rebound, a relative strategy will crash due to its net negative beta exposure.

To determine the threshold levels at any given point in time, Gulen and Petkova use prior non-overlapping return periods.  For example, at the end of December, the cumulative return from January-to-November is evaluated against the distribution of all prior January-to-November periods for all stocks.

Corey here: My interpretation is that the returns of one stock will be compared against all other stocks.  The problem, as I see it, is that this will naturally nudge the selection towards higher volatility securities.  I would be interested to see how this analysis is affected if instead of using cumulative returns, risk-adjusted cumulative returns were employed.

By using all available historical data (an expanding window technique), Gulen and Petkova create more stable thresholds for winners/losers when compared against a traditional relative strength model.  We can see that sometimes relative losers are defined by a threshold of -80% and sometimes the threshold is as high as +10%.

Absolute Momentum v Relative Momentum Threshold

 

We can see that, by the nature of the more stable threshold, absolute strength momentum and relative strength momentum will not necessarily hold the same securities.  In fact, there are times where their threshold levels are significantly different, driving to unique portfolios.

Perhaps one of the most interesting features of absolute strength momentum is its ability to navigate relative strength momentum crashes.  For example, from March 2009 to May 2009, the relative strength loser portfolio rose 156% while the relative strength winner portfolio gained only 6.5%.

However, since the absolute strength momentum strategy requires sufficient holdings in both the long and short legs, many of these crashes are avoided simply because the strategy's selectable universe is too small.  For example, in a crisis period, the winners leg rarely has enough stocks.  And without both legs holding enough securities, the strategy switches to a risk-free asset.

From January 1965 to December 2014, there are 25 months during which there are less than 30 absolute strength losers.  In those months, the average return of absolute (relative) strength momentum is 0.53% (0.45%) per month.  Over the same period, there were 27 months where there are less than 30 absolute strength winners.  In those months – the market crashes – the average return of absolute (relative) strength momentum is 0.49% (-0.73%) per month.

Absolute Momentum Number of FIrms

Absolute Strength v Relative Strength Equity Cruves

 

It could be argued that absolute strength momentum is a cousin of time-series momentum, where a security's past cumulative return is compared against a benchmark (usually a risk-free rate).  Winners and losers are defined based on whether a security is above or below the threshold.

To determine if time-series momentum explains absolute strength momentum, the authors regress time series momentum returns against absolute strength momentum returns.

They find that while absolute momentum is significantly exposed to time-series momentum, a significant intercept remains, indicating that there are unexplained return characteristics.

On the other hand, time-series momentum is also significantly exposed to absolute strength momentum and there remains no significant intercept.

Absolute Momentum Regression AnalyssiIn other words, the success of time-series momentum may actually be due to the underlying absolute momentum strategy it is exposed to.

Corey here: At this point, the authors launch into an argument regarding capital gains overhead.  I think the answer may be more mathematical: by-in-large, absolute strength momentum portfolios will be a subset of their time-series momentum peers.  This analysis seems to demonstrate that the returns driving a time-series momentum portfolio are due to stocks at the tails and not the many stocks that "just" cleared the risk-free hurdle rate.  

The results of this new study are interesting in that they (1) highlight a method for potentially avoiding momentum crashes and (2) provide evidence that time-series momentum returns may be due to returns of outliers.  

My expectation is that securities in both legs of this portfolio are highly volatile.  I would be interested in determining how this evidence changes based upon cross-sectional risk-adjusted returns or per-security generated thresholds, effectively allowing lower volatility securities into the mix.

Corey is co-founder and Chief Investment Officer of Newfound Research, a quantitative asset manager offering a suite of separately managed accounts and mutual funds. At Newfound, Corey is responsible for portfolio management, investment research, strategy development, and communication of the firm's views to clients.

Prior to offering asset management services, Newfound licensed research from the quantitative investment models developed by Corey. At peak, this research helped steer the tactical allocation decisions for upwards of $10bn.

Corey is a frequent speaker on industry panels and contributes to ETF.com, ETF Trends, and Forbes.com’s Great Speculations blog. He was named a 2014 ETF All Star by ETF.com.

Corey holds a Master of Science in Computational Finance from Carnegie Mellon University and a Bachelor of Science in Computer Science, cum laude, from Cornell University.

You can connect with Corey on LinkedIn or Twitter.