# How analytics makes football more fun

My 18-year-old daughter, Elise, has never been that interested in football. I sent her off to summer camps as a kid to try to get her interested. She played for a half season with a local team. And I took her to a few matches (most notably when she was four she saw Arsenal beat Umeå in the first leg of the Champions League final). But football wasn’t really for her. I share a lot of things with Elise, not least our temperament, but a love for the beautiful game isn’t one of them.

But now I have found the solution, in an unexpected place…

Over the last four years I have, together with Twelve football, developed and applied a machine learning algorithm for ranking football players on the basis of the actions they perform. This method helps clubs find talent and rank them on their contributions. Our method takes historical data of the position of dribbles, passes, interceptions etc. and assigns a value to those actions based on a statistical model of how these actions contibute to a team’s probability of scoring. I explain in detail how are method works in the video below as part of my mathematical modelling in football course.

There are now lots of names and variations of this type of approach (VAEP, EPV, G+, xThreat, …). Teams, most famously Liverpool, now regularly use methods like this in recruitment. Opta and Statsbomb have adapted versions of this model in their products. But I think that the simplest way to describe it is as measuring ‘the probability of a team scoring given this action is performed’, because this is what it actually does. Just as expected goals measures the probability of scoring with a particular shot, the Twelve points measures the probability of an action resulting in a goal (or preventing the opposition from scoring)

Recently, we have released the Twelve Football App. This includes a game where you pick a player and compete against your friends to see which of your players gets the most points live during the match. You can switch player at any time during the game, as much as you like. I now play with a group of friends whenever we watch a match at the same time, and it was now that I tried to get Elise interested again.

And…

She loved it! At first she found it difficult to know which player to pick. But then, after asking me which player was which on the TV, and noting how many Twelve points they were getting, she started to see the pattern. After a few matches, Elise was telling me who was best in different situations, working out who was taking and defending free-kicks and when to switch to them. Now we watch a game together once a week, and she usually beats me!

The highlight came for me last night, when Elise’s boyfriend downloaded the App and started playing. It was the first 15 minutes of Arsenal vs. Liverpool and Alex immediately chose Aubameyang. Elise just said to him, ‘thats not really a good choice, you can see that Liverpool have the ball mostly in Arsenals’ half. At this stage in the game you need to take a midfielder who makes a lot of high value passes in the middle like Thiago, who I’ve got, or maybe you can take Partey. He is defending well in those areas.’

This was a concise assessment of the state of play, which proved better than the half time analysis we later watched together, where the ‘experts’ just went on about Liverpool playing with a ‘positive intention’ (whatever that means?). The Twelve game had turned Elise in to a top-level pundit!

Well, maybe, I wouldn’t go that far. But what the experience brought home was that advanced statistics (not pointless numbers like possession and number of corners) brings a new type of viewer to football. The Twelve method goes way beyond xG (which major media outlets are now waking up to) and provides a metric that clubs use in their scouting and tactical evaluations. But the increase in sophistication, does not make the game less accessible, it makes it more inclusive. It brings a new way of watching the game.

Alex, who said he doesn’t usually watch football because ‘nothing happens’, was also hooked. But he never got close to Elise in the Twelve game. I thought I was going to beat her when I had a 20 more points than her going in to the last minutes of the match. I had taken Thiago for safety. But then Elise switched to Sadio Mane, who promptly ran up the pitch and got 0.338xG chance, giving him 338 Twelve points. Elise topped the leaderboard.

The scenes in the Sumpter household as the final points were tallied in the Twelve App were unbelievable.

Professor of Applied Mathematics. Books: The Ten Equations (2020); Outnumbered (2018); Soccermatics (2016) and Collective Animal Behavior (2010).

## More from David Sumpter

Professor of Applied Mathematics. Books: The Ten Equations (2020); Outnumbered (2018); Soccermatics (2016) and Collective Animal Behavior (2010).