I am current reading the World Cup Edition of Soccernomics written by Simon Kuper and Stefan Szymanski. These guys are “disciples” of Billy Beane, who is portrayed by Brad Pitt in the movie “Moneyball”. Soccernomics documents how only recently the football world has embraced statistical analysis. The main problem has been how to translate raw data into meaningful analysis. Many teams initially started tracking information such as passes, tackles, and kilometers covered. Unfortunately, this raw data can have little or no significance. For example, what is the relationship between total distance covered and winning? Knowing how many passes a player has made does not tell us if they are killer through-balls or square passes to a static team mate. When I started looking at some of the World Cup stats, I was surprised to see that Italian midfielder Verratti had a passing completion rate of 91% against Uruguay but then under further analysis, I saw that 33 of his 43 completed passes were to three defenders (Chiellini, Bonucci, Barzagli) and Pirlo. Counting the number of tackles can also be a poor indicator, perhaps a player in proper position is able to deal with the ball unchallenged unlike the player who arrives late and must make the tackle.
Another “lesson” from the Soccernomics book is how easily our eyes can fool us, how we can be quick to make judgements, or how we (and actually many managers and players) have pre-conceived notions, totally repudiated by stats.
One of the managers who routinely relies on stats is Arsenal’s Wenger, who has an economics degree. Wenger utilized statistics to find a little known teenage midfielder named Flamini at Olympique Marseille who consistently covered 14 km per game, to replace Patrice Vieira. OWT is also supposed to be a statistics geek and he utilizes the Uruguayan firm Kizanaro Sport Technologies to analyze La Celeste performances as well as analyze opponents and their tendencies (e.g. how they play out of their end of the field, who directs the attack, etc.).
Unfortunately at LCB, we don’t have the same tools to analyze player performance but there are a couple of sites out there that deal with stats:
1) One site is squawka.com which publishes a Player Performance Score which measures a player’s ability to positively influence a game of football. An advanced algorithm that takes every recorded on-ball action on the football pitch, evaluates the outcome, the pitch coordinates, the player position and the preceding event and provides a rating. The score is broken down into 3 components; attack, defense and possession. The attack component looks at shots, crosses, 1v1s, etc. The defense component looks at tackles, interceptions, etc. and the possession component looks at passes, through-balls. The Squawka rating can be interpreted as follows:
- Score between 10 to 20 points over 90 minutes – Average game
- Score over 50 – Player is performing very well
- Minus score – Poor game
2) Another site is whoscored.com which uses over 200 raw statistics in the calculation of a player’s rating, weighted according to their influence within the game. Every event of importance is taken into account, with a positive or negative effect on ratings weighted in relation to its area on the pitch and its outcome. The scoring system for the ratings is out of 10, starting from 6.0, with 10 being the highest score. The site actually provides live updates during games. A rating of 9-10 is excellent, from 8.0-8.9 very good, from 7.0-7.9 good, from 6.0-6.9 average, from 5.0-5.9 poor, from 4.0-4.9 very poor.
Here’s La Celeste’s player ratings during the World Cup, according to whoscored.com. Next I will look at the ratings for the latest call-ups.