The Spanish national team and Barcelona experienced close to a decade of unrivalled success, much of it based on high levels of possession and winning the ball back when not in possession. Whilst some tried to copy this style, with varying levels of success, there has been a noticeable shift to a more direct playing style in recent seasons.
Atletico Madrid were surprise winners of La Liga in 2013/14, last season Leicester City won the Premier League, whilst Real Madrid won the UEFA Champions League in 2013/14 and 2015/16, with Portugal becoming UEFA European Champions this summer, all achieving success playing a more direct style of football rather than high levels of possession football.
In this month’s newsletter SportsMatrix examined the data for the clubs in the English Premier League, La Liga and Serie A to see how levels of live ball possession and attack ratio during the season were reflected in the average number of shots per game and percentage success rate during the season.
What does percentage possession actually tell us about a team?
Does the team having higher levels of live ball possession control the game, leaving the team with less possession chasing the game?
Clearly there is no real advantage to possession for the sake of possession if a team plays passively and poses minimal attacking threat; especially if the team with less possession is the more proactive when it has the ball.
In order to see how this played out in three of Europe’s top leagues last season, we examined every league game in the English Premier League, Spain’s La Liga and Italy’s Serie A to see how possession levels and attack ratio reflected against a team’s percentage success rate and their average number of shots per game.
The percentage success rate is calculated as the percentage of actual points to the maximum available points during the season.
The attack ratio is the total number of each team’s attacks compared to those of its opponents on a per game basis, totalled for the whole season, this provides us with the average attack ratio on a team-by-team basis.We then examined the relationship between possession levels and percentage success and shots on goal and then replaced possession levels with attack ratio, to see which method provided the strongest correlation.
The table below shows the correlation of possession for the teams in each league for the season against percentage success and shots per game.
The data reveals that of the three leagues examined, there was a strong correlation between possession in Serie A and success; whilst in La Liga and the Premier League there was only a moderate correlation between levels of success and possession.
When comparing shots per game with possession, there was a moderate correlation for the Premier League, whilst in both La Liga and Serie A this correlation between possession and shots per game was stronger.
Attack ratio as a more accurate indicator
But possession data is only of limited value, since it fails to show a true picture as to which team controlled the game. So rather than examining bland possession percentages we used the value called ‘attack ratio’ to understand which team actually controlled the game.
The table below shows the correlation of percentage attack ratio for all of the teams in each league for the season against their percentage success and shots per game.
The data reveals a much stronger correlation between attack ratio and shots per game and a stronger correlation in percentage success.
By calculating attack ratio to percentage success and shots per game, we now have a clearer picture of which teams controlled the game rather than relying on possession statistics alone.
Of the three leagues examined it is the Premier League where the gap between percentage possession and attack ratio correlation is greatest, with a nearly 20% increase in percentage success and shots per game.
In conclusion we can learn from this report that in order to have a better understanding of which team controlled a game, we should definitely not be relying solely on possession statistics as a yardstick.