Thursday, October 10, 2013

I forgot to talk about Moneyball and its implications

Moneyball - the book, not the movie - is an excellent vehicle for getting at some of the issues regarding inference that we were discussing in class today and particularly on the question of whether there are systematic biases and therefore on if you understand those biases you can find "deals" in the labor market if you know where to look.  There are some things in the book that are particular to baseball, but other things generalize.  We'll consider both of those here.

First a little backdrop.  The book is now about 10 years old.  Some of the story survives intact even now.  Other parts have been severely criticized.  The book is about the Oakland Athletics and, in particular, their General Manager, Billy Beane.  He had essentially an impossible problem to solve.  The A's were and still are a low budget team, meaning they keep a payroll for the players that is in the bottom 25% or so of major league franchises.  Sometimes this is explained by saying that Oakland is a small market team.  I don't completely buy that explanation because Oakland is quite near Silicon Valley and there is a huge urban sprawl in the vicinity.  Perhaps a different and more plausible explanation is that the San Francisco Giants got there first, so they have the primary loyalty of people in the area.  In any event, ownership has pursued a modest payroll strategy.  This dates back to when Charley Finley was the owner.  He had a deserved reputation as a cheapskate.  The A's were the best team in baseball in the early 1970s.  But then free agency came along and many of the good players on the A's went elsewhere.

There is no mystery on how to keep payroll down in baseball.  That is accomplished by playing mostly young players who are not yet eligible for free agency.  Perhaps you also have a handful of veteran players, to serve as glue for the team, but the core has to be comprised of younger players who aren't paid very much (by Major League standards).   The issue is: how can you possibly win when playing mainly younger players?  Identifying top talent is difficult with lots of inherent risk.  It is on this issue where Moneyball provided two ideas that now predominate in the game.

The first regards the "high school phenom" and whether to draft such a player.  (There is an annual draft in Major League Baseball just as there is a draft in Football and Basketball.)   Finding a player in high school, or even in college, who can do well in the Major Leagues is like finding a needle in a haystack.  Such players are rare.  If the players you look at are all players who have outlier great performance in high school, they will look exceptional when compared to their competition.  The first insight in Moneyball is that some players no longer look like such outliers when they step up a level or two in the competition.  College baseball is much more competitive than high school baseball.  If a player is an outlier in College, that is a far sharper indicator that the player will do well in the Major Leagues.  The book argued that it was foolhardy to draft high school players, because it was too risky.  It is now common practice to draft college players.

The second issue relates to the role of the "talent scout" who identifies prospects early, typically based on direct observation of performance, but not in high volume.  This contrasts with doing a statistical analysis on the player's college career and making predictions based on that.  The finding is that the scouts had a mental image of the type of player they want to see.  If the player fits the image the scout would become a booster for drafting that player.  If not, the scout would not advocate for the playe.  But the scout system itself wasn't that reliable.  One could reduce the error by considering the statistical evidence.  In baseball this is called Sabermetrics, a term associated with a guy named Bill James and with Billy Beane.

Beane's personal narrative makes the story more interesting.  As an up and coming player in the Mets organization he had what the scouts were looking for.  Beane came up at the same time as Lenny Dykstra, a squat little guy who didn't fit the mold.  But Beane fizzled out and Dykstra became a star (first for the Mets, later for the Phillies).  The scouts can't observe the inner workings of the player, the passion, the flame that drives the pursuit of excellence.  Dykstra had that in immense quantities.  Beane didn't have that at all.  So the narrative suggests that Beane understood implicitly from his experience as a player that the scouts can get it very wrong, which is what prompted him to take a different approach.

There are other parts of the book that are more controversial and not so well agreed upon.  Most of the book is about identifying good hitting and what it takes to produce a team that will score runs.   Some critics of the book correctly note that at the time the A's had some great pitching and that is what allowed them to stay competitive.  There was much less int he book about using Beane's type of approach to find good, young starting pitching.

Let's now bring this back to general issues of recruiting in the labor market and biases that likely exist.  I did mention the tall thing in class, but failed to point out that it is of interest primarily because height is readily identified.  You could include other appearance variables - good looking for example - and it almost surely would correlate with pay, but we would probably have issues on how you identify what counts for good looking.  Even at a beauty pageant, the judges don't always agree with one another.

Let me move to some other factors where there might be bias.  Late bloomers may be under valued by the market.  Somebody who did poorly in classes as a freshman and sophomore may have turned it around in the junior year.  That person very well could be on the rise, but the gpa might be still rather low and not show the full potential.  Conversely, an early bloomer may have already plateaued before graduation.  In this case the GPA might be reasonably high, but future growth in the individual is less likely.  Under some metrics, this person will look like a good bet when in fact that is not the case.

Another area where the market may get things wrong is for people where English is not their first language.  They may  prove to be exceptionally good and productive people, and there English skills might very well improve, but the market may penalize them, in a similar way to how it might punish shy people.  In effect, this would say more about the provincialism of the recruiter than about anything else.

If you understand where the various biases in the market are then you might pursue a counter strategy whereby you deliberately recruit under valued people, both to identify talent and to not pay an arm and a leg for them when you've found a good person.  A belief that markets are efficient would suggest that such an approach to recruitment can't last long.  That the Oakland A's remain highly competitive, a decade after Moneyball appeared, suggests that perhaps the market is not quite as efficient as its proponents would like to have you believe.

No comments:

Post a Comment