Maths!!

Joe Devine narrates this interesting short - posted the other day on The Athletic magazine's Tifo sub-channel on Youtube - about why football is intrinsically much harder to analyse than almost all other sports.

The killer quote in the middle of it comes from Christofer Clemens, the Head Analyst of Die Mannschaft when they won the 2014 World Cup, who once noted ruefully: "We are increasingly convinced that there's a lack of data that provides real information about the things that make you successful in football..."  Even the World Cup winners didn't know how they'd done it!

The main point of the video is that a continuous sport like football is much harder to analyse effectively than one which is more broken up, divided into a series of discrete, short passages of play. In cricket or baseball, for example, there are only so many types of delivery that the bowler/pitcher can produce, and only so many ways that the batsman/batter can respond; and once the fielding side have dealt with whatever happens, the ball quickly becomes 'dead' - the game is paused while the fielding side reset and the bowler/pitcher prepares to deliver the next ball. In football, the ball becomes dead far less regularly, less often; and while some teams will try to eat time off the clock by putting the ball into touch as often as possible, it is common for sequences of uninterrupted play in a field game like football to go on for a minute or two.... and occasionally even for several minutes at a time. The possible patterns of play are thus almost infinitely complex. 

The video points out that there are a few games that are arguably even more continuous in their play: ice hockey, for example, allows for substitutions to be made without pausing the game. What it doesn't go into, however, is that these games tend to have fewer players and/or shorter playing periods - again making the challenge of analysis just that bit more manageable. Most field games have settled on 11 as the number of players; a few, such as rugby have more - but again, they usually have rules which restrict the variety of play. In rugby because the ball can only be passed by throwing, there is a fairly narrow practical limit on how far and how fast the ball can be moved with each pass; also, the ball cannot be passed forwards in that game. In football, almost any part of the body is able to be used for controlling and moving the ball, and passes in all directions are allowed, and it is possible to move the ball when struck with the foot a huge variety of different distances and at different speeds - and even to shape the trajectory of the ball by deliberately applying spin to it. This almost limitless range of potential movement for the ball also means that in football almost every player on the pitch can potentially receive the ball next, whereas in most similar sports there are generally only a very small number of likely receivers. (Stick-and-ball games like hockey have similar fluidity of movement, but are much more heavily biased towards passing the ball in a particular arc of the field.) Thus, even a relatively short passage of play, just 4 or 5 passes, can move the ball enormous distances and involve almost the entire team. There are few other games - probably NONE - which have anything like this level of variety and complexity in their potential patterns of play.

The video notes that as statistical analysis has become more thorough in recent years, it has started to prove useful in some areas - particularly in identifying the 'skills profile' of individual players, and deciding if they might be a good match for a given squad and style of play (Brighton's business model is founded on this, and they've become very, very good at it). Where it still falls down, and probably always will, is in assessing the effectiveness of tactical approaches, and determining how far they contribute to a team's overall success.

I would suggest that this is not just because it requires one to look at an entire game (or a whole series of them) rather than just individual game actions, and at the team as a whole rather than just an individual player - although obviously that is a massive (and insuperable) part of the problem. Surely, it's also that it's impossible to define with any precision or consistency what team tactice are. They change from minute-to-minute, as players seek to adapt to shifting circumstances on the field; they change as the manager makes in-game tweaks, or galvanises his team to greater efforts with a rousing half-time pep talk; they change with shifting game-states. They change from game to game, as a manager seeks to adapt to a particular opponent. They change from season to season, and even within the course of a season, as managers seek to stay fresh - and surprising to the opposition. And however clear and consistent the tactics may be in the manager's vision, how cohesively and consistently and effectively they are realised on pitch by the players can vary drastically from game to game.... and within a game, sometimes even from minute to minute. Plus, of course, the effectiveness of the chosen tactics always depends not only on how accurately the team impements them, but on the response of the opposition. 

In a fluid game, like football, with an almost unlimited range of possible movement, and long uninterrupted sequences of play.... any attempt to statisticallly analyse success or failure in terms of the overall team tactics is usually going to be doomed to failure. 

Caveat: Unless the tactics are really, really bad! If a team has a really obvious flaw, like Ten Hag's United having such a huge gap between their midfield and defence, because they didn't have the pace in defence to mount a high defensive line, even though their high-pressing style really required that - then, yes, you can see why a team is losing all the time, and the statistical data will expose that flaw too. But when you have two very good teams going up against each other, it's far more daunting to try to disentangle the impact of the tactics from the importance of individual moments of skill from the players. [Maybe Pep's tactics have never been that good, and he's just been saved over and over again by the brilliance of his teams...??]

I'd go further and say that I think you can usually determine the tactical basis of a result in indvidual games via the 'eye test' - watching closely, and analysing multiple different interacting factors at once. It's much harder, usually almost impossible, to reliably draw such conclusions from data alone. And disentangling the impact of 'tactics' - amid the web of other elements: form, fitness, confidence, refereeing decisions, the 'luck of the bounce', the quality of the opposition - in overall results over a run of games.... that, i think, is a very elusive - probably illusory - Grail to be chasing with statistical analysis.

Food for thought.