Book reviews Week: Moneyball
Moneyball by Michael Lewis
Published by W. W. Norton
Unless you’ve spent the last year living under a rock or in Andy Carroll’s armpit, you’re sure to have heard of Moneyball. Either you’ve read the seminal Michael Lewis book, seen the Brad Pitt film adaptation, or read one of the torrent of blogs, articles, and sundry internet scoffings about the statistics based system developed by General Manager Billy Beane for acquiring undervalued talent for his Oakland A’s Major League Baseball team. For the six or seven people who aren’t familiar, here is a gross simplification:
Moneyball the book chronicled Beane’s replacement of traditional ‘eye test’ scouting of players with a system of ‘sabermetrics’ reminiscent (and possibly inspired by) of the methods employed by the pioneering amateur analyst Bill James. This system identified a correlation between a hitherto underused stat (‘On Base Percentage’, or the percentage of the time that a given batter would get to first base or beyond, including walks) and runs scored, and the subsequent correlation between runs scored and winning percentage. Beane found that players with a good On Base Percentage were more important to a team winning than those with better stats in the ‘pure’ batting stat categories, and better fielders – but since nobody else had worked this out, players with good On Base Percentage came cheap. He was therefore able to assemble an extremely low budget, but hyper-efficient team that had a great deal of success in the regular season whilst never quite putting together a run to the championship.
So far, so blah blah blah. What’s this got to do with football? It’s a pertinent question. Some ten years after Michael Lewis’ book came out, the terms ‘moneyball’ and ‘sabermetrics’ began to creep into the blogosphere and even the mainstream press. Billy Beane himself said that his methods could and should work in football. A cynic might attribute this to the appearance of Brad Pitt’s film at the beginning of this year – but Damien Comolli’s resignation from Liverpool in April 2012 put stats firmly and legitimately on the menu of both the professional and amateur hack.
By this point, moneyball and ‘sabermetrics’ had expanded in meaning, becoming shorthand for a player acquisition strategy that was over reliant on statistics. Comolli had quite openly based the purchases of Jose Enrique and, infamously, Stewart Downing at least in part on their impressive stats for ground covered, forward passes made, and chances created. Sod’s law was particularly strong in 2011/12, as Downing recorded one of the most statistically pathetic seasons since geeks first began stopwatching time of possession. Comolli’s stats based approach quickly became a stick to beat him and owner John W. Henry with, and article after article appeared to bust Liverpool’s reliance on the strategy.
Excellent points were made about why the approach failed in this case. Most point to the truism that statistics cannot tell the full story of a football match, because, unlike baseball or cricket, each “play” does not have a comparatively small set of concrete outcomes. Each at bat in baseball must end with the fielder on base, scoring, or out, whereas in football, once a ball lands at a player’s feet the number of options and results feels almost infinite by comparison – perhaps a little more infinite if it drops to Andres Iniesta than if it plonks onto Matt Mills’ size tens. The upshot of this is that it’s hard to find the “secret stat” – the undervalued number that both correlates to your team winning more games, and is undervalued by the competition. The Daily Mail (of all places) even ran a feature suggesting that the key stat was the number of short, intense sprints a player was able to make in 90 minutes.
Another suggestion was that Liverpool had not properly applied the moneyball method – that they had forgotten that the most important part of the doctrine is that your player purchases need to be undervalued by your competition, so that you can assemble an efficient squad and squeeze the most out of every pound. As Dave Whelan could tell you, the best way to get value in the transfer market is not to buy British – and certainly not to spend £35 million on raw English strikers with half a decent season behind them.
But it could be said that none of this is really moneyball in the first place. The use of stats to evaluate players is nothing new – Opta have been around since 1996, and messers Allardyce and Wenger, among many others, have been credited with sound use for years – but moneyball seemed more involved, more pseudo-scientific, and unforgivably for some, more American. Would it be unduly suspicious to suggest that saying you’re using moneyball strategies could just be a convenient way to make fans, media and gullible owners think there’s something more advanced going on with a club’s transfer policy than there really is? Liverpool are the highest profile adopters of the way of the stat, but they are by no means the only ones – but at least they have a clear approach to chuck out and move on from. They’re moving on.
For me, moneyball philosophy at its heart isn’t really about stats. What drove Billy Beane was the thrill of winning against rivals who had greater financial brawn, but considerably less brain. He loved to acquire a player asset for cheap, pump up their value, and sell on to other general managers desperate to emulate his success. Buy low, sell high. If it sounds like financial trading, that’s because that’s exactly where the whole concept comes from. Moneyball is about smaller budget teams identifying something, anything, to let them compete with, and beat, the big guys. The mechanism for Billy Beane was statistics, but in football it seems clear that that isn’t the answer. Instead, clubs have to find other ways of playing, other places to find their players, new tactical and organisational systems, not the equivalent of On Base Percentage.
This has never been more pertinent in football than now, with the mega-rich adding a zero to their income and their outgoings every year.