The Twittersphere was buzzing yesterday, and the question everyone was asking was: how can Andy Carroll be worth £35m? The general consensus was that he wasn’t worth that amount of money and that the footballing world had gone crazy.
Maybe the conventional wisdom is correct. But one thing “Moneyball” (in my opinion the best Sports Business book ever written) taught us is to ignore conventional wisdom. Moneyball is about the Oakland A’s baseball team and their General Manager Billy Beane and is currently being turned into a Brad Pitt/Philip Seymour Hoffman movie, due for release later this year:
The basic premise is this: In 2002, the Oakland A’s had the second smallest budget in baseball (around $40m) – less than a quarter of the New York Yankees ($126m). But for the previous 3 years, they had consistently been one of the top four teams (though they hadn’t won a World Series). That is the equivalent of Wolves qualifying for the Champions League for 3 years in a row on their current budget (the A’s didn’t increase their relative budget as a result of their success in the first two years).
The secret to the A’s success was to completely re-think the way they evaluated players. Using a new set of statistical analyses (called Sabermetrics) and throwing out all conventional wisdom, they were able to see that the market for players was hugely inefficient. Some player attributes were highly overrated in terms of their correlation with success, while others were highly undervalued. So the key to running a successful team on a budget was simple:
sell those players who have overrated attributes (for lots of money) and buy players who have the underrated attributes (for much less)
Well first and foremost, John W Henry, the new owner of Liverpool is a convert to Sabermetrics. Many of the techniques used in Sabermetrics came from the financial markets, which is Henry’s background. And when Henry bought the Boston Red Sox in 2003, his first move was to offer Billy Beane a job for a guaranteed $12.5m over 5 years (which he turned down). Nevertheless, he installed Sabermetrics at the Red Sox, who then went on to win multiple World Series.
So Liverpool is now a Sabermetric club with the best brains in the business analysing players’ value. Would they really pay £35m for a player if they didn’t see the value?
And what did that analysis look like?
Success in football is defined by winning points. Given the financial rewards at stake (eg. qualifying for the Champions League), it is relatively easy to calculate the value of each Premier League point. And if points are the asset, then goals are the currency. To quote from Moneyball:
“Before the 2002 season, Paul DePodesta (the A’s sabermatrician) had reduced the coming six months to a maths problem. He judged how many wins it would take to get into the play-offs: 95. He then calculated how many more runs the Oakland A’s would need to score than they allowed to win 95 games: 135. (the idea that there was a stable relationship between season run totals and season wins was another Jamesean (the father of Sabermetrics) discovery)”
It is possible to determine how many goals you need to score in order to acquire your targeted number of points. And therefore, each goal has a value. Before we have any Ossie Ardiles and Kevin Keegan arguments about teams who score lots of goals but don’t necessarily win points, here are the facts:
1) Currently, the top 4 teams in the table are the four teams who have scored the most goals
2) In the seasons 2006/2007, 2007/2008 and 2008/2009 the top four teams in the table were also the teams who scored the most goals
3) The only exception to this rule in the recent past is last season, where Manchester City, who finished 5th, scored more goals than Tottenham, who finished 4th
So, each goal that Carroll contributes can be valued in terms of ‘acquiring’ points.
But how do we determine how many goals he is likely to contribute?
One of the cleverest bits of analysis discussed in Moneyball is the disentangling of the link between what actually happens on the pitch and what is expected to happen. To quote again:
“Any ball hit any place on a baseball field had been hit just that way thousands of times before: the average of all those hits was the Platonic Idea (of an average run value). Call it a line drive that is hit at x trajectory and y speed to point #968. From 10 years worth of data, you can see that there have been 8,642 practically identical hits. You can see that 92% of the time the hit went for a double, 4% for a single and 4% it was caught. Suppose the average value of that event is .50 of a run scored. No matter what actually happened, the system credits the hitter with having generated .50 of a run…”
So let’s apply that to football and Andy Carroll.
The first thing we do is to forget the number of goals he scores and the number of assists he makes (what actually happened) and concentrate on the expected value of his actions on the pitch.
Here are a few examples (by the way, these are all completely made up assumptions but it wouldn’t be too difficult to calculate them):
1) A penalty is scored 79% of the time. Every penalty he wins is worth .79 of a goal (regardless of whether he takes it or even whether it is scored)
2) A goal results from a corner kick 4% of the time. Every corner kick he wins is worth .04 of a goal
3) Divide the pitch up into sectors and calculate the percentage of time a goal is scored from a free kick taken from that sector. If a goal is scored 7% of the time a free kick is taken from sector 4, then every free kick Carroll wins in sector 4 is worth .07 of a goal (and the same for all the other sectors on the pitch)
4) Divide up the goal into 6 sectors (top left, top middle, top right, bottom left, bottom middle and bottom right) and calculate the expected goal value of a shot on target in each of those sectors. If 46% of shots into the top left sector are goals, then award him .46 of a goal for every shot he hits into that sector
5) A successful pass within the opponent’s penalty area results in a goal 8% of the time. Every successful pass he makes in the penalty area gets .08 of a goal
6) Winning a header in the opponents penalty area results in a goal 5% of the time. He is awarded .05 of a goal whenever he wins a header in the opponents area
…and so on
From this type of analysis, we can calculate the expected number of goals Carroll will contribute to Liverpool and we will also know the value of each goal. If that is more than £35m over the course of his contract, then it could represent great value.
Of course, this has been simplified a little bit. The real analysis is the ‘marginal’ impact of Carroll compared to another striker. In other words, how many more goals would he be responsible for than another striker?
And this is where the idea of undervalued attributes comes into play. For the sake of argument, let’s assume that “winning headers in the opponent’s penalty area” is worth more in terms of expected goals than people give it credit for. If Carroll is the “Greek God” of winning headers in the opponent’s penalty area, then he is worth much more than people think he is.
Would we really be surprised if John W Henry’s Sabermatricians have been doing this type of analysis since they bought Liverpool? Maybe they have just sold a player for £50m who has ‘overrated attributes’ and bought one who, even at £35m, is undervalued.
By Carsten Thode on February 1st, 2011