24 July 2011

Football by Numbers (Or Why Luka Modric is a Great Player, No Matter What You Say)

It is little wonder that Luka Modric is amongst the most coveted players in this summer’s transfer window. He is now a player who appears destined to feature in the canon of great Croatian players, alongside figures such as Partizan Belgrade legend Stjepan Bobek and volatile midfielder Zvonimir Boban. But here is an interesting statistic for you: since Modric joined Spurs, they have conceded fewer, scored more and won more games when he did not feature in the side. It sounds so counterfactual and yet it is true. What, if anything, are we to conclude from this? Has Modric’s purported talent been confounded by the statistics? Can it be the case that we were wrong in our appraisal of him? Harry Redknapp’s mind boggles.

It is not too anachronistic to claim that over the previous decade there has been a greater interest in the gathering of football knowledge. This relates not only to the discussion of football history and tactics, but also match data. It may be a banal point to make, but clearly extensive television coverage and the internet (particularly the role of blogs such as the superlative Zonal Marking have played an important role in this change. Football need not be confined to the pub; it can be found in the coffee shop or even, dare I say it, the library.

A corollary of this glut of statistics in the modern game is a trend, in certain circles, towards numerophilia. This is not limited to fans and journalists, but is evidenced within clubs too. A recent article by Simon Kuper has detailed how the data revolution in football has been accompanied by developments in football’s quantitative methodology. The article comments on the many ways in which stats are acting as drivers in the game, contributing to (and in some cases determining) transfer policies, tactics and player ratings. Such an approach can greatly assist clubs in terms of identifying players that are suitable for them based on any number of desired variables (for example, on their chance completion rate). It can also have a positive impact on the coaching of teams; Kuper offers the example of Arsène Wenger substituting Dennis Bergkamp late in matches after noting that his output tended to decrease around 70 minutes into matches. A millennial eye forward might even anticipate stats as powerful as GPA (Goal Probability Added) figures, which would detail the added probability of a goal arising in correlation with the inclusion of a given player.

Long-time stat fan

This is all interesting enough, but there is a problem with the use of statistics that is occasionally overlooked (if not by the professionals then at least by their acolytes). All experience is interpretive. More specifically, raw data is only as useful as the lens through which it is interpreted. If you do not know how to make the data legible then it is at best of limited use and at worse violently misleading. How, for instance, should we determine which stats make a good team? Which stats are indicative of a good forward? What is the factor that demonstrates that a team is in charge of a match? A good example is offered in Kuper’s article. We might assume that tackling is an important aspect for a defender and, in many cases, this holds true. Yet Paolo Maldini, one of the greatest defenders of all time, only made, on average, one tackle every two games. The statistics used and the significance attached to any given reading of them imparts their intelligibility and meaning. And the potential misreading of statistics does not solely apply to the rating of players; it can apply to any aspect of the game in which there is interpretable data available to hand. An example of this is provided in David Runciman’s review in the London Review of Books of Scorecasting: The Hidden Influences Behind How Sports Are Played And Games Are Won. In this book Tobias Moskowitz and Jon Wertheim defer to the stats in every case. Runciman focuses on their explanation of the benefit of home advantage and why this arises. They dismiss factors such as the support of the home fans, the inconvenience of travel and local knowledge, leaving only one remaining factor: referees are far more likely to favour home teams and give decisions their way due to the atmosphere in the home stadium.  Runciman describes this as a ‘lovely theory’ and one that fits in nicely with many of our standing assumptions. But it isn’t true. Deconstructing the argument offered in Scorecasting, Runciman puts the selective use of statistics under scrutiny in order to demonstrate how they could be read in other ways if we wanted to support a rival hypothesis. For Runciman the problem must be left open rather than prematurely closed and catalogued as a known fact. In reference to Jose Mourinho‘s outstanding home record he notes that ‘the idea home advantage can be reduced to, an blamed on, referees is just the sort of conservative, risk averse thinking that gets you into trouble. So [Mourinho] knows something the freakonomists don’t: you can’t always trust the numbers.’

The refusal to prematurely close an idea for the sake of its cataloguing as a known fact is reminiscent of the epistemology of the silly Nazi, Martin Heidegger (whose irritatingly opaque - albeit very good - work, I’m sorry to say, I will now discuss in brief). As Mary Jane Rubeinstein has noted, ‘Heidegger claims it is his task not to think the totality of thought as the self-unfolding of being, but rather to think that which has not been thought as the self-withholding of being.’[1] For Heidegger truth (or what we have been referring to as the facts) should be understood as unconcealment or emergence from hiddenness. This conception of truth can be contrasted to the Leibnizian notion of the ‘adequation of reality and thought’, where truth has to present itself to a representing subject. This representing subject, or monad, is man and all being is measured against its representability by man. For Heidegger this line of thinking is demonstrative of the tendency of beings (beings of being) to put themselves before being (the being of beings). Rubenstein again: ‘With escalating fervour, the Western philosophical tradition has codified its “objective truths” at the expense of truth itself, covering over every absence of presence and every mystery with the certainty of full representation. By attempting to incorporate everything in its path, however, philosophy has only pushed the unassailable event of being and/as truth farther away, thereby severing from being the very epistemological subjects and objects it purports to secure.’[2] This is not to say that calculative-representational thinking cannot be effective. The success of prediction using statistics, for example, is a testament to this. But since ‘calculation cannot calculate unconcealment, it risks shutting itself off from truth itself.’[3] It is curiosity that drives man forward on this quest for the categorisation of facts. But this restless urge to catalogue under the call of Gestell (the desire to enframe in a representable form) only takes us further from the truth and does not allow being to be on its own accord. This, perhaps, is the key point: letting being be – something that calculative-representational desire does not allow.

Martin: Post-war and grumpy

Now, this stance is very easy to interpret in the wrong way. I am not suggesting that these numbers and statistics are meaningless. Heidegger too admits the effectiveness of calculative-representational thinking and there is clearly a lot that you can interpret from match data – to ignore it would be nothing other than Luddism. Rather my point is that this data (the stats/numbers) would be subject to the truth, not vice versa. Such an understanding accords the appropriate dignity to statistics whilst also respecting what is most real – the game itself and the truth inherent to it: the concealed unconcealment.

What I have advocated, then, is taking stats with a pinch of salt. It is not that they are of no use, but rather that they are just one method amongst many of arriving at astute conclusions. The debate, Kuper surmises, is between jocks and nerds, between the old fashioned manager relying on gut instinct and the modern quant with his spreadsheets and pie charts. It is interesting that Kuper presents the dichotomy in this way: after all, what is more muscular than a statistic? The language of the statisticians can be just as obstinate and final as the language of those relying on instinct. Kuper quotes Oakland A’s general manager Billy Beane saying, ‘What stats allow you to do is not take things at face value.’ Of course in one sense this is true, but is it not also the case that stats replace one face value with another? Statistics are neither more nor less inherently violent than gut instinct. Either can present itself as a last word. Either can lead to false judgement. It seems that the real debate has been obfuscated. Whether or not one relies on statistics is not the deciding factor in how good a manager one is. Rather it is what one does with the numbers, or what one’s instincts suggest – it is the extent to which one knows the truth (for instance, that Berbatov is better than Hernandez). A case in point: Kuper cites Real Madrid’s mistake in selling Makelélé to Chelsea as an example of typical gut reaction. But this is not fair. It was, rather, an example of a bad gut reaction in the context of a bad transfer policy (the Galactico project). There were plenty of people at the time (including some at Real Madrid) who, relying on their own gut feeling, understood Makelélé’s worth and viewed his sale as a mistake. To reiterate: the determining dichotomy here is not between gut and stats, it is between good decisions and bad ones, good managers and bad ones.

This has been a long way round to saying something fairly commonsensical, but at least it allows us to say it. It allows us to say that we should not dwell too much on the stats. It allows us to say that Modric is Tottenham’s best player, even if they often do better without him.

Harry Redknapp fears what he does not understand

[1] Rubenstein, Mary Jane (2008) Strange Wonder: The Closure of Metaphysics and the Opening of Awe, New York, Columbia University Press, p.25.
[2] Ibid, pp.26-7.
[3] Ibid, p.27.


  1. One thing one might want to point out is that Modric may have only played against better teams (or would have not played against weaker teams), a pattern seen with many top players. I don't know if this is the case, but if it is, then there's an alternate reason as to why Spurs score more/allow fewer when he's not in the side.

    One has to utilize statistics *correctly*, also.

  2. I only recently found the blog, by the way, and it's wonderfully well-written. Definitely bookmarked on my browser!