What if you let loose a data scientist on in-match decisions?

What are some very silly ways that you could use very advanced data to make in-match decisions in football?

Each year that Get Goalside has been running when Christmas has rolled around I've sent out newsletters that try to max out the silly while also being interesting.

In 2019 it was a broad one, 'The silliest things you can do with advanced football tech'. Last year it was broad in a different way, 'Pick a sport, any sport, and take ideas from it'. So this year we're gonna go specific: what are the stupidest in-game changes you could make with very advanced use of data.


You could go about this two ways. The first is to work out when time-wasting will help you; the mental model we all have probably works fine, but there might be some times when the data surprises you. The big factors are going to be score, proximity to the end of the game, and a little bit about how well each team is playing.

However, the more fun way to use data for this would be to help change your style of play to make time-wasting easier. Are your long balls going straight through to the goalkeeper? Is the opposition left-back slower to take throw-ins than the opposition right-back? Would it be better to clear a cross up-field or smash it out of play into the crowd? Instead of building a model whose ultimate aim is goal-scoring, build a model whose ultimate aim is wasting the most time.

Data can make the art of time-wasting even more elegant.

Designated tactical foulers

There are teams who definitely seem to rotate 'tactical foul' duties, and it's not like it's a complicated system: "I just fouled someone, can you foul next time buddy?". But teams are getting the hang of data nowadays, so why not throw some numbers at the calculation?

If, as a manager, your team is being countered against, you could definitely use tracking data (which tracks every player as they move around) to see which of your own players is close to the counters. Could you use this information to put them in better positions to stop the attacks in the first place? Sure! But you could also pair that with some kind of 'likelihood of getting booked' model and, like, tell a full-back to invert and the central midfielder to stay wide instead because one of them has a history of more convincing cynical fouls or something.

Fouling more

On that theme...

All of the 3D images we've been getting of offsides in the Champions League and World Cup this season are produced with skeletal data. Lots of high-quality cameras all around the stadium have their images run through sophisticated software to know where every player's ankle joints are.

If you had some NASA-grade data people, I'm sure you could build some kind of system that highlights when a given amount of contact between two players would usually be called as a foul. With this, you might be able to say "hey, the data says this ref isn't calling fouls - or more specifically isn't calling this type of foul - we can afford to be a bit more aggressive."

I feel bad for referees now, but that would be a fun use of smart people.

Football Index Insider Trading

This is the one thing that makes me sad that the Football Index bubble burst.

If you're bored of building statistical models that tell you which players increase or decrease the chances of winning (who isn't!) why not pair that with "whose fake gameified stock value will increase the most if we win this game/start involving them in the game more/designate them as the penalty taker".

It would be morally wrong for football clubs/coaches to have been trading Football Index-type player stocks anyway, but if you're gonna be shady then you may as well do it right.

Sprinkler settings at half-time

At the end of the day, everything on a football pitch is a matter of physics. Newtonian forces and crap like that. Let's say you're a team who like playing passes in behind the opposition defence, and in the first half they've been bouncing too high for the forwards to bring under control. That's physics: the interaction of the ball and its speed with the firmness and dryness of the grass.

So maybe you could run those passes through one of your fancy physics models but you change some of the variables based on the pitch surface moisture level. Does the bounce deaden a little if the grass is wetter? How much pitch moisture would be too much before the passes just skid through to the goalkeeper? Adjust the sprinklers at half-time accordingly.

Would it make a meaningful difference? I highly doubt it. Would the story being leaked cause a great couple of days for fans to talk about? Absolutely.

Create leading survey questions to fool your web3 NFT-token-holding members into thinking they're taking part in a genuine democratic process

It's not something that really happens at the moment, but it's conceivable that, at some point, a bunch of crypto/web3-type clubs are going to try and give their members the option of weighing in on in-game decisions.

You may, like certain conservative-leaning political parties in certain countries of geopolitical significance, decide that too much democracy is a dangerous thing. In which case, why not work out what you think you want to happen, and then get some kind of Natural Language Processing system to write the survey questions and answers in such a way that it primes the voters to vote in a certain way.

Would it work? Probably not enough for it to be reliable but just enough to get you banned from being involved in running a football club ever again.

Conduct the crowd

I haven't yet read any research about the effect of geeing up the crowd on how a team plays, and in fairness it would be difficult to get the data, but it's the kind of research I think I want to read most.

Imagine your winger's low on confidence. They're not having a good game. You don't know how to change things tactically. But wait: the decibelometer indicates that the section of crowd that the winger is playing near are quieter than usual by two standard deviations!

You send word out that that needs to change. Luckily your intern has done a research project on home season ticket holders and their atmosphere network effects. You know the exact seat numbers most likely to start and sustain a chant. You contact your crowd management agent in their section and get them to start a loud and positive discussion of the winger and how they need a boost. The chant starts. The winger's head is held high. Three minutes later they go on a mazy, full-pitch dribble and score.

In another life you'd be overthrowing governments with this set of skills and support staff. But here, the only leaders you're deposing are your rivals at the top of the table.