The world has to thank Italy for lots of things. Food: pasta, pizza etc; Fashion: Gucci, Armani etc; Fast cars: Ferrari, Maserati etc. Then there’s the ancient history, the gondolas, and the jauntily-angled tower. This brief cultural tour has so far overlooked at least two other spheres in which Italians have had a hefty influence. There is, of course, football: Baldini, Buffon etc; and, of all the categories to bring up, political thinkers: Machiavelli, Gramsci etc. Being a big fan of tenuous linkage, it’s the marrying of these latter two categories – football and political science – where this blog post begins. For the decidedly apolitical, please bear with me!
Giovanni Sartori is a name unknown to most except the most ardent of politicos. As far as those with an interest in political party systems go, however, Signore Sartori can be classified alongside Machiavelli and Gramsci as yet another Italian to have made an eminent contribution to the political literature. To classify states by their party system, Sartori created a typology that numbered seven mutually exclusive categories, ranging from “one party” (think the uncontested and omnipotent CCP in present day China) to “extreme pluralism” and “atomized” systems (hypothetical possibilities wherein a state would have a mass of parties all with little power). Apparently not one for hyperbole, Sartori described his typology as self-evidently innovative, and to an extent it was. Before Sartori, there were those who offered a simple binary “one party vs multi-party” dichotomy and others who rejected classification altogether on the basis that attempts to divide up a continuum were perceived to be folly.
Now we come to football. As regular readers of this blog and followers of my tweets will know, the general analysis of goalkeepers is thin indeed. What research has been done is mostly focused, perhaps understandably, on shot-stopping. Other aspects of keeping, such as distribution, have been almost entirely overlooked. Distribution is a difficult nut to crack, and presently finds itself in a place not dissimilar to party system analysis pre-Sartori.
There are three options presently available to the amateur analyst: 1) Simplistic analyses can be performed that look at pass accuracy% and average pass length as Flavio Fusi did in his attempt to create goalkeeper radars, with such data available at Squawka for numerous goalkeepers in numerous leagues. 2) Alternatively, hours at a time can be spent manually copying data from Statszone where the pitch can be divided into thirds to assess keeper accuracy (see below screenshots of Petr Cech’s passing in last weekend’s North London derby). 3) The final option is to consider passing lengths as a continuum and thus impossible to fairly analyse. This is also the easiest and least time consuming option by virtue that the analyst can throw their hands up in the air and go and do something else.
None of these options are without problems. Option 1) is a methodology riddled with problems, as the distance data favours goalkeepers who like to kick long (or who play for managers who tell them to kick long), and the pass accuracy% favours those doing the exact opposite (plainly it is easier to complete a short 2 yard pass to an unmarked defender than it is to lump a long-ball into an opposition penalty box stacked with defenders). Option 2), whilst better than option 1), also makes some key assumptions regarding the homogeneity of average pass data within the thirds, whilst option 3) is flawed in that distribution would remain a facet of goalkeeping worthy of only qualitative analysis rather than the all-important triangulation.
It would seem, then, that an attempt at option 2 would be the best avenue to pursue. The issue regarding the homogeneity of the thirds, however, is fairly major. The following graphic evidences the lack of unity in the average pass accuracy% within the thirds:
And this issue manifests itself latitudinally as well as longitudinally:
A satisfactory way in which to categorize goalkeeper passes is thus presently lacking. Whilst a simple solution might be to divide the pitch up into multiple sectors, say 9×9 or 9×3, analysts would then be presented with another problem: small sample size.
If only Sartori was a football analyst…