A version of this blog originally appeared on the Newfound Research blog site.
At the very core of our risk management and relative-performance capture models is the philosophy of momentum; specifically, we use a proprietary spin called dynamic, volatility-adjusted momentum. There are many possible explanations for the momentum anomaly. Personally, I ascribe to the behavioral finance explanation: irrational behavior and positive feedback loops drive asset prices beyond the explanation of rational valuation.
While we normally point to the persistence of momentum across asset classes, geography, and history as evidence that it must be an innate factor within market participants, I came across an interesting New York Times article from 2007 about a study that exhibited momentum in a very different field. The article “Is Justin Timberlake a Product of Cumulative Advantage?” describes a Columbia University experiment performed online with 14,000 participants. The purpose of the experiment was to explore the independence of taste and preference in music selection.
Participants were asked to explore, listen to, and rate music. One participant group would be able to see how many times a song was downloaded and how other participants rated it; the other group would not be able to see downloads or ratings. The group that had “social influence” was then sub-divided into eight distinct, random groups where members of each sub-group could only see the download and ratings statistics of their sub-group peers.
The hypothesis of the experiment was that “good music” should garner the same amount of market share regardless of the existence of social influence; hits should be hits. Secondly, the same hits should be hits across all independent social influence groups. What the study found was dramatically different. Each social-influence group had its own hit songs, each of which commanded a much larger market share of votes and downloads than songs in the socially-independent group. In other words, downloads and votes exhibited much greater kurtosis when social influence was a factor.
Introducing social influence had profound consequences: it made hits bigger and it made hits more unpredictable. The author called this effect ‘cumulative advantage’ — I call it momentum. As summarized in the article, ‘Because the long-run success of a song depends so sensitively on the decisions of a few early-arriving individuals, whose choices are subsequently amplified and eventually locked in by the cumulative-advantage process, and because the particular individuals who play this important role are chosen randomly and may make different decisions from one moment to the next, the resulting unpredictability is inherent to the nature of the market. It cannot be eliminated either by accumulating more information — about people or songs — or by developing fancier prediction algorithms, any more than you can repeatedly roll sixes no matter how carefully you try to throw the die.’ Similarly, as cited in the article, technology products exhibit a “network effect,” where the attractiveness of a technology is a function of its social acceptance. Early traction can lead to market domination.
If we consider that our preferences can be modified by social influence — and that our preferences can actually become reflective of the market — then we can see the massive snowball effect that even a modest amount of injected randomness can have. For this very reason, we utilize momentum at the core of our investment technology instead of trying to predict, because the slightest variations can have profound impacts on the future. No amount of algorithmic complexity can correct for this fact. We believe it is better to be reactive and adaptive to changing markets and align our models with human behavior rather than swim against the very, very powerful current of social influence.