The big debate in the game of baseball over the past decade or so has been somewhat crudely summarised as 'stats versus scouts'. New forms of statistical analysis, as exemplified and brought to a wider audience by Michael Lewis' Moneyball, have gradually begun to take hold within the game, and organisations are slowly but surely incorporating this into their day-to-day activities. It is no surprise that the more progressive organisations in this respect have also been successful in recent years, notably the Boston Red Sox and Cleveland Indians. Stats versus scouts is something of a false dichotomy, of course: the key issue to running a successful baseball franchise is using the correct (or non-traditional) statistics in conjunction with traditional scouting to give a fuller picture when evaluating talent and making decisions.
Moneyball focused upon the General Manager of the Oakland Atheltics, one Billy Beane, and his ability to run a small market team successfully. This he did by acquiring players with skillsets not recognised by traditional scouting methods or by traditional statistical methods, or to put it another way, in finding talent which was incorrectly undervalued within the market. Such thinking was viewed as counter-intuitive within many baseball hierarchies at the time, but it was ultimately successful, as the A's continually challenged for the playoffs ahead of teams with significantly bigger budgets.
As I have mentioned on this blog before, I came to take a deep interest in baseball around the year 2000, right as this statistical revolution was taking off. The spiritual father of this movement was Bill James, who had been writing about the flaws inherent in traditional methods of evaluating baseball since the late 1970s, however, it took the advent of the internet before this debate could truly flourish. James himself was hired as a special consultant for the Boston Red Sox in 2003.
At this time I always wondered in the back of my mind about how such thinking could be potentially applied to my true love, football. What is striking about the advent of sabermetric thought (and which explains a lot of the resistence to it from 'traditionalists') is that it is so rational. Using sabermetrics does not necessarily entail using complex statistical methods to crunch numbers (although this does exist), rather, it is about viewing problems rationally and thinking about them objectively. To put it another way, it involves stripping away a lot of the mystique inherent in being a sports obsessive and putting events in their correct context. Could this be done in football?
Lets look at two basic examples from baseball. Batting average (hits per at bat) and RBIs (runs batted in) have been the bread and butter forms of gauging the 'good' from the 'bad' since baseball began. However, they are inherently flawed means of doing so, and do not really tell us how effective a player is in helping his team win, or how good he is at his job. Batting average is flawed when compared with On Base Percentage (OBP), which factors in walks, and literally tells us how many times a player can get on base (rather than just getting a hit). Getting on base is the single most important task set to an offensive player in the game of baseball. Similarly, RBIs are flawed in that a player is dependent on his team-mates being in a position for him to 'drive them in.' It does not demonstrate his adeptness at being a productive offensive contributor in the larger sense.
I am rather sceptical when considering the possibilities of applying statistical analysis to the game of football, for the simple reason that baseball is a static game with a limited number of possibilities in any given scenario (for example, there can be no outs, one out, or two outs; there can be no runners on the bases, one runner, two runners, or three; etc.). It can be broken down and compartmentalised in ways that football quite simply cannot. It is quantifiable almost in its very essence. However, that is not to say that it strikes me that many means which are currently used to evaluate footballers are flawed.
Clean Sheets: Simply put, clean sheets are not the definitive measure of what makes a goalkeeper good (or, by extension, an individual defender). Clean sheets are context specific, as are most actions which take place in a football game. Unsurprisingly, the best teams are most likely to keep the most clean sheets, because they have better defenders, keep possession more, and as a result find their own goals threatened less. Does this mean that the goalkeeper is deserving of more praise for simply playing behind a good team?
In theory, a goalkeeper could have zero saves to make while still being credited with a clean sheet. Conversely, a goalkeeper could find his goal bombarded for the duration of a game, make a succession of very good saves, yet still end up conceding four or five through no fault of his own. Luck can play a huge role. For what it's worth, clean sheets are one of six criteria used by Actim in their player ratings, which they use to calculate 'the value of each player's contribution.' Simply put, I think clean sheets can be a good measure of team, but not individual performance.
Passing: Pass completion and total passes have been used as a gauge of efficiency, especially for midfield players. There is a problem with this, which can be summed up by the following: all passes are not created equally. For example, a holding midfielder who is deployed in front of the back four (think Hargreaves or Makelele) can be assumed to have both a higher pass completion rate than a player playing higher up the pitch, or one who is deployed to be creative (think Riquelme, for argument's sake). The holding midfielder's job is to win the ball, and give to to someone who can play. Roy Keane was reknowned for always playing the simple ball and Didier Deschamps was derided as a mere 'water-carrier' (the implication being that the position was easy to play and merely required legwork). By definition, their job is to make simple passes. This is also true of defenders.
Conversely, further up the pitch when we take creative midfield players or attackers, it can be reasonably assumed that they will have a lower pass completion rate, because as they draw closer to the opposition's goal, the relative importance of a forward pass must surely increase. It becomes more of a high-leverage situation. And surely the opposite is true, namely, that pass completion in defensive situations cannot be compared to that in offensive situations.
Again, lets not forget the context of the game. Are passes completed when a team is stroking the ball around the back four and the midfield while getting into the game in the early going, or going through the motions while three up with ten minutes to play, really as important as those completed when the game is in the balance? To put it very simply, it seems crude to compare all passes as being somehow equal in this way, either through pass completion or total passes made.
Assists: Assists are still valued as a major offensive category (check Actim or your fantasy league), and while I do not dispute that quite often an assists (ie a pass leading directly to a goal) can be decisive, it would be wrong to assume that assists are the most decisive play leading to the scoring of a goal. Often they can merely be quirks of circumstance. Perhaps the decisive individual play involved in the scoring of a goal was a tackle to win the ball and set the play in motion, or perhaps it was the momentum given by a pass which itself came four or five passes before the 'assist' leading to the goal. The football pitch, it would seem, simply provides too many possibilities in the way that the baseball diamond, with its limited number of outcomes, does not. Or so it would seem.
My point is this: many methods which football fans use to try to quantify the game are flawed. However, football does not lend itself to the sort of analysis which has so enriched baseball. Certainly, it is not as prominent, but it does seem to be going on. For example, see the American based Matchanalysis.com, who have a number of German Bundesliga and MLS sides already using their analytic methods. They have an interesting approach, using video analysis to break down the game into a series of individual plays. How do they utilise this? Well, you need to sign up to find out...
Lets bring this full circle. Billy Beane, the star of Moneyball, and still general manager in Oakland, is something of a football buff (a Spurs fan, apparently). Beane has given a number of seminars to Premiership managers and coaches regarding different approaches to talent evaluation. He is also part of the ownership group which is resuscitating the San Jose Earthquakes team in the MLS. Beane intends to take a hands-on approach, and has commented that "I don't pretend to have any answers .... I'm just hungry to find them."
It will be interesting to keep an eye on the San Jose experiment, but one feels that the application of more innovative methods of talent evaluation really need to be tested in the big European leagues. How viable is it really? Only time can tell.
