The Statistical Problem With Soccer

The outcome of the game is never meant to be a "true reflection" of the teams' abilities, but a reflection of their ability against that side, on that day, in those conditions. A round-robin style tournament is used to have a team playing different opponents, a mix of different styles. Central attacking sides. Defense and counter-attack. Wide open matches. Closed central movement. Teams alter their strategy against specific opponents and try to find holes in their game and exploit an edge.

This seems like a problem with the statistical model, not with the sport, because "A > B > C > A" is possible in a team sport, and not just because of a luck factor. One team may have strengths that exploit certain weaknesses in other teams. That are uncountable examples - Team A lacks fitness and is weaker in the last 20 minutes; but this will only be exploited by a team who have a high fitness level. Team B has as weak left wing, but this will only be exploited by a team with a strong right winger. The list goes on.

Sports popular in the U.S. almost seem designed to allow for statistical analysis. Baseball is the ultimate game for a statistician, as you have data for every single pitch thrown and the outcome. American Football breaks the game down into series of discrete events (downs) each of which has a numerical outcomes (yards gained or lost). Basketball generates 10s of events every game with a definite outcome that can generally be assigned to a single player (shot made, missed, turnover, free throw, rebound, etc.).

All of this gives fans ways to analyze and rank individual players and teams in a fairly objective way. Thus, the immense popularity of fantasy leagues.

Soccer is inherently more subjective, as there are far fewer events with an objective outcome. With scoring a rare event, and no obvious ways to assign numerical values to all the things that happen between scoring chances, subjective analysis is all that's available.

Which ties back into the original article in that freak occurrences and dumb luck can have a much larger impact in soccer. One team can play much better for most of the game, but a freak goal, controversial call or uncharacteristic error can have an outsize impact in a low scoring affair. This is mitigated in the round robin format rounds with accumulated wins and points, but is very much a factor in one-and-done games.

I believe I read before that at one time even the final round was a multi team affair, with the champion determined by cumulative totals against the other teams, but it was done away with because fans wanted a definitive championship match between just two teams.

Treating it as "an experiment to determine which of the teams is the best" is not a good approach. There are a thousand contests in 90 minutes, but probably only two or three of them result in a goal, one of which may be from a penalty anyway.

High scoring games are much better at determining the best team on the day.

I believe that one of the major reasons for soccers massive worldwide spectator popularity is the low scoreline. You watch the game even if your team is obviously outclassed, as you can still win, and you can still win with just a few minutes left in most cases.

There are 32 teams playing the World Cup, if one of them is a 1:3.5 dog, they have a pretty DAMN GOOD chance. The betting odd is usually sth like 1:4 or 1:5, I think.

More importantly, the team which wins is the better team, by definition. It is a question whether they can perform consistently in a series of matches with the correct strategy each time, whether they can adapt to changes, whether they are focused at the important moments. Certainly there will be luck and drama, but the team which plays better (for most of the match duration) wins much more often than that number 28% tells you.

Look at professional poker. You can for sure say that there is more luck involved, but times and again the familiar faces show up at the final tables.

It's a game, it's meant for fun, it's not a science experiment.

By having it semi-random, you spread out then ups and downs.

Someone did an analysis of video games (posted on slashdot, but I can't find it), where they showed that the best video games don't have solid excitement all the way, but rather ups and downs.

The same it true here. If your team always won you would be bored pretty quickly.

1. It's mainly a knockout tournament which obviously increases the variance

2. International sides aren't a cohesive team like domestic club sides only playing a handful of times per year leading up to the tournaments.

Compare this to the domestic leagues where the league format reduces the variance and players spend most of the year working with who they're playing with on the pitch. In the EPL, the #1 ranked side going into the season probably wins the league 75% of the time. It's a much lower number though for the domestic cups because of the knockout format.

"People don't gamble on Chess."

So very quickly:

> Their approach is to think of a soccer game as an experiment to determine which of two teams is the best. The question then is this: what is the probability that the outcome of the experiment truly represents the relative abilities of the two teams. And the answer, unfortunately, is not very probable.

This is not a bug; it is a feature. Spectator sports have drama and theatrics and exist first as entertainment, and only second as a contest of skill.

In Major League Baseball, the New York Yankees are far and away the best team this year. It's not even close. Almost all their position players are between top 3 and top 10 at their position, they have a very solid top of the rotation and mid rotation, and good to amazing relief pitching.

They'll compete with eight other teams to win baseball's championship, the World Series. I'd put their chances around 22% of winning it all. But they're far and away, clearly, the best team in baseball.

But that's why people watch. People don't gamble on Chess.

A beat B, B beat C, A loses to C should tell you that there most be something sophisticated going that your model isn't capturing.

American Football fans will surely remember when Vince Young was starting his career with the Titans and all the commentators had this standard line "He just wins games!" because his team was winning despite his terrible numbers.

This analysis doesn't strike me as any more substantial than that ridiculous phrase.

Another problem is that soccer really has a level of randomness that is considered a feature of the sport. We don't really expect that the best team (in technical terms) always win. In Brazil, people commonly say that "soccer is a little box full of surprises".

http://arxiv.org/abs/0909.4555

The end results do seem to favor the best teams.

The only people who don't recognize that are England fans, who routinely overestimate their team anyway.