All New Wilmott Jobs Board                     (g)

Rooney Will Score Because Best Did

On page 97 of Nassim Taleb´s best-selling probabilistic masterpiece “Fooled By Randonmess” (hardback, first edition) there appear these illuminating ruminations: “In the beginning, when I knew close to nothing about econometrics, I wondered whether the time series reflecting the activity of people now dead or retired should matter for predicting the future. Econometricians who knew a lot more than I did about these matters asked no such question; this hinted that it was in all likelihood a stupid inquiry...I am now convinced that, perhaps, most of econometrics could be useless – much of what financial statisticians know would not be worth knowing”.

Thoughtful stuff, no doubt. And polemic too. If anything, the field of “financial econometrics” seems stronger than ever, with prominent academics and academic institutions devoting lots of attention to it, and with one of its inventors actually receiving the Nobel prize just a few years ago. And yet, it is hard not to agree with Taleb. When it comes to the financial markets econometric analysis is likely to be less relevant than usually assumed in theoretical circles.

At its core, econometrics is an attempt to forecast the future based on what happened in the past. As every former and present economics student worldwide can attest, this exercise can involve extremely complex statistical and mathematical maneuvers. Lately, econometrics has found its way into financial research, giving birth to the relatively newish discipline of financial econometrics. Past market data is used to predict future market movements, through the use of funky models with increasingly funkier names. But intelligently designed as these tools surely are, it is not easy to become a believer. Simple old-fashioned common sense ruthlessly dictates that past information should not be very useful in preciting the future of financial markets.

Why? Mostly because, as Taleb very originally points out, we would be trying to predict what current financial players are going to do based on what ancient players did in the past. In the markets, prices move for one reason only: human action. Clearly, each human being has their own, independent, decision-making capabilities. The financial prices of a certain historical period would be the result of the actions taken by those individuals active in the market at that time. Those prices thus reflect the average consensual decisions of the players that happened to be around, given the relevant circumstances then present.

Econometricians would try to use those prices to forecast the prices of several periods later (say, today). The problem is that many of those individuals originally involved in setting the prices included in the time series used in the analysis would by now be either dead or no longer active in the market. Econometricians would in fact be borrowing from inactive brains, attempting to predict the decision-making process of a group of independently-thinking individuals from the decision-making processes of a different group of independently-thinking individuals that are no longer around. Why should Peter´s particular stock pickings twenty years ago matter for predicting Paul´s particular stock pickings today, particularly since Peter has been retired in the Bahamas for the last decade? It might be sensible to use data from Peter´s past actions to predict Peter´s current actions, but it definitely looks a bit suspect to use that data as a predictor for the actions of another, different, unrelated human being.

What financial econometricians are trying to do seems akin to predicting the number of goals to be scored by a football club next season by looking at the time series of goals historically scored by that club. Just like financial prices are the product of specific actions by very specific humans, goals at any one time are the direct result of the actions taken by those players then present in the pitch. As any football fan would tell you, it would be weird to try to infer anything relevant from goals-scored data that includes players that no longer play. There are simply different people involved. Since goals (like stock, bond, or commodity prices) are all about people it seems far-fetched to assume that historical time series can tell us much about the future, no matter how complex the techniques involved.

The fact that the legendary George Best managed to score 180 goals in his time at Manchester United in the 1960s tells us nothing about the scoring capabilities of today´s striker Wayne Rooney. Similarly, using past data to predict future financial prices seems like an exercise in reviving the dead.