The winter break in the Dutch football season is still in full effect. The downside of this is that for at least another week there are none of the usual match reports to bring here. But the advantage may be that such a period offers an opportunity to sit back and contemplate on the process of analyzing football matches.
The current article is a result of that. It focuses on the rather simple question of how to measure a team’s performance in a football match, and more specifically, how to determine each individual player’s contribution to his team’s performance? A simple question that brings about a lot of thought.
Conventional measures of individual player performance are quite limited. Ever since there are football competitions, top scorers charts have been around, and in more recent times, sometimes assists charts too. But other than that, there are hardly any standards to compare player’s performance by. And what do top scorer charts say?
Imagine striker A playing for team X and striker B playing for team Y. Both teams play each other and striker A scores the winning goal in a 1-0 match. Along the current standards he is likely to be credited with a good (and measurable) performance, while his counterpart, striker B, is unlikely to come away unnoticed in the long run. And what about the midfielder who completed around one hundred passes to give his team a dominance of possession in the first place, or the defender making that last ditch tackle to save his team from going 2-0 down?
Imagine that in this particular match striker A, who scored the winning goal, also missed several good goal scoring chances while striker B won his team a penalty that was missed by a team mate. This might put things into a different perspective and of course this information will find his way into match reports the next day. But a few days later, striker A is the one who made the scoring charts and his missed chances will be less remembered (i.e. not recorded), while striker B, who did everything right except perhaps not taking the penalty himself, drowns in anonymity.
With the aid of a tool measuring the strikers’ performance based on recording single actions instead of ‘end parameters’ like goals, the strikers’ individual level of play would be recorded in a much more honest fashion. Striker A would be credited for scoring the goal, but would lose credit for missing a series of other opportunities, and striker B would be credited for winning a penalty.
For almost two years now the Castrol Rankings website offers a rather sophisticated points rating system, awarding players from five competitions around the world (EPL, Serie A, Bundesliga, Primera Division and Ligue 1) as well as from the Champions League, points per match based on individual actions during the match derived from OPTA data. The players’ match actions are compared to a league average for his position on the field and points are awarded accordingly. Points scored for the entire spectrum of match actions are added into one figure, probably with the use of a weighting factor in order to represent the different levels of importance of the respective match actions. The end result is a single rating per player per match that should allow for an objective score of individual player performance.
However, there are severe objections and limitations to the current format. First and foremost, the exact formula’s with which the match rating is computed is not known, so it remains difficult to interpret the resulting end score. Furthermore, the match rating is adjusted on the basis of the ‘importance of the match at hand’ and the ‘level of competition the match takes place in’. The correction factors for that, again, remain unclear.
Besides other, relatively minor, limitations there is one more serious objection to mention. The results are presented on a 12-months rolling average basis, causing a serious delay in the player rating lists. For example the persistent high ranking of Wayne Rooney, long after his level of play had dropped seriously, made a lot of users realize the limitations of the current format.
At present , these rankings are the only form of objective analysis of individual player’s performances available. Unfortunately the rankings are presented as a single value incorporating the complete match performance, rather than allowing users to compare different aspects of their game. The explanations offered in response to user’s questions on FIFA’s World Cup page help to understand that different aspects of the game contribute to the overall score, but without any insight into how players perform in these different aspects, the practical use of the score systems remains limited.
Despite all these limitations, let’s not forget that it is a major step forward to see intelligent use of statistics being introduced to football. After all, in other sports these have proven to be very influential. Take, for example, the innovative effect that the amateur statistical analysts of Football Outsiders have had on both tactics and journalism in American Football. And recall Oakland Athletic’s GM Billy Beane from Michael Lewis’ outstanding book ‘Moneyball’, who says he prefers to watch as little baseball action as possible in order not to let the randomness of his personal observations get in the way of dealing with the objective data regarding his players’ performances.
To compare individual player’s contribution to an average standard has the potential to offer football fans a lot more understanding of their team’s performance. Think for yourself, do you know how many goal scoring opportunities your team scores from the average corner? Or what rate of shots from outside the box your goalkeeper has saved this year? Point taken?