• Momentum strategies are popular with investors seeking long-term returns
  • But the strategies have experienced a rollercoaster run since the advent of COVID-19
  • An approach combining good and bad momentum components could be the answer

Last year proved a remarkable one for momentum strategies. The tech giants that had done so well in recent years continued to outperform at the beginning of 2020, and the same high momentum names were further buoyed by investors’ preference for the digital over the physical economy with the onset of COVID-19.

As the world wrestled with a deepening public health crisis, growth assets continued to outperform value until momentum experienced a dra- matic reversal at the start of Novem- ber 2020. Following positive news on the effectiveness of a COVID-19 vaccine on 9 November, investors quickly sought cheap distressed assets, triggering one of the largest momentum crashes of the past 20 years.

The early months of 2021 com- pounded this shift, with expectations of rising rates and investor optimism triggering a reversal in appetite from high-growth to value hurting the momentum trade still further. Moreover, after some momentum strategies developed value exposure, growth came back into favour in 2021 as investors faced an uncertain trajectory for the pandemic.

The proverbial rollercoaster experienced by momentum in 2020 is reflected in the 40.4% return on a long-short momentum portfolio of global developed stocks for the first 10 months of the year, followed by a dramatic fall of 23.8% in November.

Strategies with exposure to momentum would have experienced significant losses in just one month as the factor unwound in November.

The volatility of momentum strategies, including the possibility of momentum crashes, has been well documented. Despite the risks, however, the strategy has received, and continues to receive, a great deal of interest in investment literature, as it is regularly identified as having a significant relationship to future returns.

The most common definition of a stock’s momentum is return over the last 12 months, usually with the most recent month excluded. The momentum factor is known to have different relations to future returns when defined over different time frames; short-term momentum has, on average, a negative relationship with future returns (mean reverting), while one-year momentum is positively correlated with expected performance (mean averting).

Instead of thinking of momen- tum in terms of return timeframe, we prefer an alternative approach to momentum in which returns-based signals are divided into two components. These are: returns on stocks that can be supported by the underlying characteristics of the company in question – ‘good momentum’; and a second part that cannot be explained either by recent news or company fundamentals – ‘bad momentum’.

Good-bad momentum versus standard momentum-linked returns

Good momentum tends to mean avert irrespective of the timeframe over which it is measured, while bad momentum tends to mean revert, whether it is measured over short or longer periods.

Not only are good and bad mo- mentum positively and negatively related to returns, respectively, but also momentum’s behaviour over different timeframes can be explained by the simple sum of good and bad momentum. Over the short term, bad momentum dominates, while over the long term the returns that relate to company fundamentals tend to persist.

It is straightforward to combine these two components to produce a stock selection signal that is simultaneously long good momentum and short bad momentum.

This composite signal has two key advantages over standard momentum. First, it adds the benefits of shorting a reversal signal to the portion of return that tends to persist into the future. As such, it tends to identify stocks with returns supported by good fundamen- tals that have also exhibited recent and unexplained reductions in price.

Second, the signal has little exposure to momentum risk, as it is constructed by buying and selling two different components that are both positively correlated with momentum. That is, the momentum exposure of good momentum is hedged by the shorting of bad momentum to reduce the overall momentum exposure of the final signal.

We have studied the performance of good minus bad momentum through the momentum crash of March 2009 following the global financial crisis and the momentum crash of 2020 (see figure).

The portfolio constructed using the good minus bad composite signal not only weathers the junk rally of 2009 but also avoids the magnitude of drawdown experienced by momentum last November. Moreover, the good minus bad portfolio also exhibits comparable positive returns to the momentum portfolio from the begin- ning of 2020 to the end of October of the same year. That is, while the momentum portfolio gives almost all its positive performance back over the month of November, the good minus bad momentum composite proves a strong stock selection signal throughout most, if not all, of 2020.

Further, the good minus bad composite signal is able to avoid some of the pitfalls of a naïve momentum strategy without any need to estimate the level of momentum risk inherent in the momentum strategy. It is derived solely from components of return and can be constructed once returns have been observed.

It is particularly interesting that the good minus bad momentum composite was able to weather the two big downturns of the past 15 years. While the global forces that precipitated both market events were markedly dissimilar, the composite signal was equally unaffected in both cases, as the sensitivities of good momentum to the dominant return drivers were offset by the opposing sensitivities of bad momentum.

By immunising the return-based signal to momentum exposure, it is possible to create a signal that not only produces positive returns when win- ners continue to outperform but also tends not to participate in the value destruction of momentum crashes.

Edward Rackham is co-director of research at Los Angeles Capital