Uma caixinha de surpresas: como o componente aleatório afeta a tomada de decisões no futebol

Abstract

There is a well-known saying that “football is a box of surprises”, suggesting that the sport presents many surprising results. Such results would be, in part, the result of random factors that would have a relatively large influence on football. Based on this premise, the article's main objective is to criticize the evaluation of performance in football based solely on the analysis of match results, which seems to happen not only among fans and commentators, but among club management bodies as well. As a secondary objective, the work seeks also to show that the random component in football results is more relevant than in other sports and, subsidiarily, seeks to provide, in an introductory manner, alternatives that help improve the decision-making process in sport. To evaluate the relevance of the random component in different sports, ten seasons were selected from 7 important football, basketball, volleyball and handball leagues (in different countries). The occurrence of surprising results (victories of the last placed teams in matches against the first placed teams) was analyzed and also the ability of bookmakers to adequately model the results of the matches. The descriptive data analysis method followed an exploratory approach, with tests of differences in population means for unpaired data and with unknown and different variances, when appropriate. The analyzes conducted effectively showed a greater relevance of the random component in football in comparison to other sports, which means that performance evaluation based only on observed results is suboptimal, because in addition to the systematic component captured (which alone would provide a good measure of performance), you can also get a good deal of random noise. If performance evaluation based solely on results is suboptimal, we need to look for ways to improve the decision-making process. We could start by looking at the expected goals metric – which, we argue, would better reflect team performance rather than goals scored (which effectively make up results, but which are more affected by random factors). We could also look for the factors that influence the creation or avoidance of expected goals, thus trying to maximize the systematic component in the formation of results (even if we cannot completely eliminate the random component). We therefore maintain that evaluating the process – and not the result – should become the way any club operates. Because it is possible to have a good process, but a sequence of bad results; just as it is possible to have a bad process and a sequence of good results. But in the long run, the greatest probability is that good processes will lead to good results and bad processes will lead to bad results.