re: New soccer book using sabermetrics

I don't know if Soccernomics was intended to be anything other than a cursory introduction of a relatively new way of understanding the sport. It's certainly not thorough, and I don't think it pretends to be. It's a light read in the style of pop social science that tries to make dense subject matter simple, interesting and engaging for the masses.

That said, I definitely do think there is a place for statistical analysis in the sport, but it needs to be contextualized so that it has meaning. The real problem with the way we read stats right now is that they help us understand why events happen after they happen; they don't seem to be able to predict future events with a greater accuracy than before, at least to my knowledge. Someone out there will figure out a way of distilling that knowledge, but they will have to account for a wide variety of variables, from pitch condition to weather to tactics to formation to style, etc. In that sense the data of the future might read something like: Team A's chance creation rate increases when player X moves into quadrant Z, etc, etc. If over the course of a season you develop a metric of events (for lack of a better word) which more or less happen in the same fashion, you can determine what variables are important and give them more weight.

I think soccer has a game complexity that is equal to chess, because of the wide variety of options a player has when he is with the ball. I'm sure someone good at math could develop the specific number for how many options a soccer player has. So let's say that someone develops a number similar to chess's Shannon number. In this sense you could say that the moves in a soccer game always tend toward originality, as the longer the game is played, the more likely it is that an event will occur which has no precedent (I think in chess they call this an "out of the box" moment).

In soccer, unlike chess, a player isn't limited by directional movement, thus he is free to move the ball anywhere he pleases, theoretically. In a realistic sense, a player is limited by some very specific options.

Let's say a LB wins the ball from a RW, and attempts to recycle play. His most likely options are to clear the ball down the line toward the opposition zone, or recycle the ball, play out of the back, and help move the ball up the field in a manner which minimizes the chance that possession is lost. While these are his most likely options, I, the observer, am only guessing based on prior experience watching other LB's in similar positions behave. He could do the unwise thing and attempt to dribble out of the back, but that seems statistically improbable. So he passes back to the GK, who moves the ball to the RB. Now the options that are presented to the RB are entirely dependent on the formation of the team, the tactics of the moment, the style of the team, the abilities of the players in his team, the abilities of the players of the opposing team, the weather, the pitch condition, the score, the time of the game, his position on the pitch, his team's position on the table/place in the cup/place in the tie. All these variables, and some I probably haven't accounted for, can play into the decision he makes with the ball, or they can be completely ignored. If one was thorough enough, you could account for every decision in a match this way, with the hope that you could glean some number or algorithm that would reduce the complexity of the game to a manageable number, and thus make the statistic which corresponds directly to a specific event, unrestrained in its application.

I hope this makes sense.