The silliest things you can do with advanced football tech

It’s Christmastime for those celebrating it [semantically, is it also Christmastime if you don’t?], and so I needed a newsletter topic that I could write well in advance. A recent conversation led to me having such a silly idea that I knew, just knew, that I needed to run with it.

So here it is: the most utterly silly things that you could do (in theory) with the most advanced tech and data in football.

Merry Christmas, happy holidays, or just have a nice December day. And if you enjoy this, please give it a festive share.

Which player runs the weirdest?

(using body pose technology)

This is the idea I had to kick it off, although I’m not entirely sure how I managed to get to it. ‘Body pose’ technology takes in video of sport (it started off in basketball, but is starting to be applied to football too) and essentially produces a stick figure of each player. And now you have shedloads more data to mess around with!

The main uses of body pose that are being talked about are knowing which direction players are facing at any given time, because that’s not in many (or any) datasets at the moment. You could also use it for other genuinely useful things, like scientifically studying a player’s technique of a given skill.

But you could also, presumably, get very very big computers to analyse how every player runs, and then run that through another very very big computer to see who the statistical outliers are. Is Raheem Sterling’s Velma-esque running really so unique? Is Jordan Henderson’s gait, as discussed by Sir Alex Ferguson, really so different to that of other modern footballers?

Now we’d be able to tell.

Who’s the best ballkid?

(using optical tracking data)

This is a combination of two of the most on-trend things in (different circles of) football right now (as of writing, and may have already dated).

In one sphere, José Mourinho and Duncan Ferguson have led to a massive increase in the number of Premier League managers hugging ballboys (granted, from a non-existent starting point). In another sphere, there are a rash of companies promising the ability to get ‘tracking data’ — knowing where players are on the pitch at any millisecond in the match — from cameras placed around the stadium. Some are even getting it from broadcast footage, requiring just one camera angle.

It all works by training software to identify which blobs of humanoid-shaped colour are football players, and so presumably it could do it to identify ballkids sitting at the side of the pitch as well.

And if it can pick up data on where the ballkids are and what their movements are, then it can give results on how long they take to get the ball back in play, which could be pretty cool. Useless, but cool.

This is where what stats nerds call ‘score effects’ come in though. If a home team’s losing, then they’ll want the ball back quickly; if they’re ahead, they’d prefer to dawdle. This would need to be factored in to any ballkid rankings.

(NB: I wouldn’t be surprised at all if there have been clubs in the past who’ve done basic analysis on this kind of thing, but this would take it to the next level).

Who are the best tactical foulers?

(using pitch control models)

‘Pitch control’ models are, like, sooo 2018, but they could be a really important building block for quantifying a football match. In short, you calculate — based on things like ball position, player positions, player trajectories — how much control each team has of each area of the pitch. (Image taken from this blog post from the ever-intriguing Barcelona Innovation Hub)

An image of a pitch control model in action
An image of a pitch control model in action

With these models, one is able to calculate who has the balance of control in any part of the pitch.

Now, teams who are being counter-attacked will often commit tactical fouls somewhere in midfield before the counter has time to develop. My idea is that tactical fouls made when the defending team’s control in their own half is weakest are the highest-value tactical fouls. If the defending team has high control in their own half when they make a tactical foul, it’s probably a waste.

(Let me know if I’m wrong, but this seems like the most tangible benefit of a pitch control model based on the (low) current levels of public knowledge about them)

Which goalkeepers are the most needlessly showy?

(using tracking data and body pose technology)

Simple. We all know that some goalkeepers, sometimes, pull out a save for the cameras.

With a mixture of tracking data and body pose technology, you could work out how far goalkeepers had to move to make a save, the time they had to make up that distance, and therefore which saves feature unnecessary flourish.

Who are the best at wasting time when they get substituted?

(using GPS/tracking data)

This is the least high-tech, given that players have been wearing GPS units in their shirts for a number of years now. They generally keep the data in-house for various reasons, apart from number of sprints and kilometres covered being released.

But why stop there!? Why not tell us the average speed of timewasting substitutions? And who are the worst, who’s letting the side down when they need to be timewasting? And which referees are the best at hurrying up players in skooting them off the pitch?

The public demands an answer.

Who gives the best hugs?

(using ‘wearables’)

Wearable technology is what it sounds like: it’s all the technology that players can wear. At the moment it’s mainly those sports bra GPS sensors, but boot trackers are coming in too.

With so many sensors all over the place, and with a sensor unit already strapped to the small of players’ backs, it surely can’t be too difficult to add another chip in there and measure data on how players and coaches hug.

Of course, for this you’d need to collect data on how the objectively best huggers hug, so presumably Jurgen Klopp would have a busy couple of days. But I’m sure nobody would mind too much.