Researchers Develop A System That Predicts Soccer Goals
Football (or as some call it Soccer) is very unpredictable. In this week’s Champions League fixtures, Chelsea F.C. was up against Paris Saint-Germaine (PSG) F.C. at their home ground Stamford Bridge. Despite having the advantage of home ground, and playing against 10-men PSG team after their main striker Zlatan Ibrahimović was red carded, Chelsea was not able to go through the round of 16 as the match concluded 2-2 and PSG went through on away goals rule.
Similarly, FIFA World Cup 2014’s semi-final played between Germany and Brazil is another classic example of this game’s unpredictability. Germany thrashed Brazil 7-1 but if you look at number of shots both teams had, you would be shocked to know that Brazil had more shots on goal than Germany (18 vs 14 shots and 13 vs 12 shots on goal).
The Disney Research scientists looked into this problem of unpredictability and tried to find out patterns that result in a team’s win or loss. The researchers team published the paper, “Quality vs Quantity: Improved Shot Prediction in Soccer using Strategic Features from Spatiotemporal Data” at the 9th Annual MIT Sloan Sports Analytics Conference in Boston, on February 27th.
According to their findings, more chances don’t necessarily mean more goals. The phase of the game — corner, counter-attack, free-kick, penalty, open-play, set-piece — is also important in determining whether the team or player would score or not. In addition, they also found that strategic features such as interaction of surrounding players, speed of play, and shot location play an important role in determining the match’s score.
They used data of season’s worth of shots on goal from an anonymous professional league. It was obtained from sports analytics company Prozone. The data included a ten-second window of play before each of the 9732 shots were taken.
Patrick Lucey, the research team lead, said that the role players play at a specific time during the match depends on the context of the match as much as their preassigned duties. By clustering each play into a specific match-contexts, the team found that all chances are not created equal. This answers the question many managers have in their mind: why a player doesn’t score despite having so many chances?
They further mentioned in the paper that highest percentage of goals resulted from counter-attacks (is this the reason behind Real Madrid’s good goal-scoring record?). After counter-attacks, most goals were scored from set pieces from a cross from a free-kick, with corners getting third highest percentage, and open play and free-kicks getting fourth and fifth spot respectively.
These statistics can help managers in better understanding their team and analyzing the game beforehand. Furthermore, people who are into bets can safely know before the match which team is going to win.