It’s obvious that professional soccer clubs suffer significant decreases in performance when players become injured, and in many cases failure to win competitions or avoid relegation has been directly attributed to unfortunate losses of players at key times. Leading edge work by Svexa’s Nick Harding and Anna Cruse describes a way to explicitly quantify the cost of individual injuries in terms of points lost and final finishing positions. This contrasts with a standard approach of simply reporting the salary paid to injured players. In advance of the full publication of their work later this year, we find out more about the background of their work.
Q: What was the impetus for this project?
A: Many of us here at Svexa are long time football fans, indeed we started our company after several top tier clubs asked for our help optimizing their use of player performance data. Watching regular coverage of games, it’s a given that key players contribute to season performance, but the extent to which this happens is a bit of a mystery. We thought it would be a really worthwhile question to answer, “what if a club had to go without their key player (or players) for several months of a season”, i.e. what would the effect on their points total or finishing position be?
Q: Can we use player salaries as a proxy for team injury cost?
A: The total salary paid to injured players is a reasonable proxy for the severity of the injuries suffered. However, salary is an imperfect indicator of ability - for a variety of reasons, but the main issue is that it doesn’t capture the effect on team performance. We need to consider other factors, such as squad depth and the strength of individual positions. This analysis enables us to put hard numbers in terms of points and league placings on injuries, which is of value to help decide how much resource to invest in injury prevention, or squad reinforcement.
Q: So what did you do?
A: We gathered data from a variety of sources including player abilities, positional preferences, their tenure in the team etc. In addition, we used external effects such as stadium size, how many games they played recently, home or away, and so on. And, crucially, we included known injuries to determine ‘squad strength’. With all those ‘inputs’, using a statistical model we estimated probabilities for the W/L/D outcome of every game and used these to simulate entire seasons many times over.
Q: What did you find?
A: Our simulations can assess what impact a certain player’s injury likely had on their club’s possible outcomes for the season overall. We do this by producing a distribution of possible results for each club in a league, with and without the specific injury. Of course, soccer is a stochastic game, and there is a large amount of variance in the results due to random chance. However by making many thousands of simulations, we can estimate the average cost of each injury in terms of points and league position lost.
Q: What can clubs do with this work?
With this sort of simulation, clubs will be able to gauge the true cost of injured spells for each specific player. Then they can appropriately scale their investments both in player purchases and in the training and recovery methods used to avoid injuries.
Q: What’s next for your analysis?
A: We are working on a series of vignettes that apply the model to a number of different scenarios, allowing us to place estimates of direct costs on real world situations. And we are expanding the analysis to multiple European leagues.
Nick is Director of Data Sciences at Svexa. He holds a PhD in health modeling and is Principal Bioinformatician at the University of Oxford.
Anna is a Performance Scientist & Data Analyst at Svexa. She holds a M.S. and B.S. in Kinesiology and Sports Science, and is a former U.S. International rower and coach.
If you'd like to find out more about Svexa's work with elite soccer teams, contact us at firstname.lastname@example.org