White Paper Combining Momentum and Mean Reversion in a Short-term Trading Model
Mean reversion refers to investor’s tendencyto overreact to new market information andtherefore create prices, which are either too lowor too high with respect to underlying securityvalue. Assets then revert to the mean price of acertain period.
Even though WFM is showing momentum characteristics, investors also overreacted at certain times and the stock "reverted to the mean".
A good amount of research has been carriedout to characterize the short-term nature of theeffect. Antweiler and Frank (2006) concluded that“overreaction to news is the typical pattern on themain American stock markets between 1973 and2001”. The authors observed timeframes of up to30 days after the news event. Others, such asLehmann (1988) also found "sharp evidence ofmarket inefﬁciency in the form of systematictendencies for current 'winners' and 'losers' in oneweek to experiencesizable return reversalsover the subsequentweek in a way thatreﬂect apparentarbitrage proﬁts".
If there is indeed asigniﬁcant mean reverting effect in certain assetprices, it would be possible to trade against thenews, sell into market rallies and produceabnormal returns. However, research is quitecontroversial on the strength of the effect. CXOAdvisory Group (2011) for example investigatedmean reversion characteristics of a major indexand could ﬁnd “only a weak reversion over shortintervals”. It seems like the anomaly is moreevident on an individual security than on an indexlevel and in general weaker than the momentumeffect.
combining both effects
Both strategies have their advantages andchallenges when applied in a clear-cut fashion:momentum-based investing requires certainmarket environments and can produce significantportfolio drawdowns when momentum stocks fallout of favor. Mean reversion type trading on theother hand can create nice steady returns, butcan also suffer from so called black swan risks:since these strategies often prohibit the use ofstop levels, a single non-mean reverting eventcan seriously damage performance. A solutioncan be to combine both approaches.
Conceptually, the combination is fairly simplesince both anomalies act on different time frames:the strategy is to trade mean reverting moves ofsecurities, which have shown certain momentumcharacteristics. In other words, momentumdefines stock selection, mean-reversion definesentries and exits.
In the past, trading costs had been prohibitiveof short-term trading strategies for most individualinvestors. With the rise of direct access brokersand their ultra-low commissions in recent years,mean reversion trading has become feasible froma cost standpoint.
Investors tend tooverdo it during theirbuying frenzies orselling panics.