It's a new year in the Gregorian calendar, the year 2022 of the common era. What a wonderful collection of almost-binary digits that is.
2021 can be captured in this excellent blog post of links from Jan Van Haaren (some posts from this newsletter are featured there, so it must be a good post).
But we are now into 2022. And here's what I think is in store.
Possession value models are to everything else in a football match what xG is to shots. They quantify things that happen on the pitch. But quantifying anything and everything else is far harder than quantifying the value of shots.
As such, analytics is all over the place. Hell, we don't even have one name for the thing yet.
I think there are two factors that'll make 2022 a big year for possession value models. The first is just that it's another year that people have had to work on them and another year of conference papers. The second is that I think theory is gonna help and that more people are thinking about the theory.
Expected goals were easy. When players take a shot there's only one thing they're trying to do, and there's a very limited number of situations that they can come from. Phases of play only matter inasmuchas it affects pressure on the shooter and a clear sight of goal, things which some data sources are already including.
A (small) theme in 2021 was an emphasis on data engineering being important in the practice of analytics. 2022 will be the year that starts to get really drilled into the public analytics consciousness.
One of the reasons for this is that more people are getting more jobs in the professional club space. But another, I think, is that if any areas of the media are looking at The Athletic and considering doing data work, then they're also going to need some engineering to set the data up right.
Heck, it may just be a year of data-as-tech. Tech solutions are nothing new in football, and 'data/analytics companies' have been doing tech offerings for as long as they've been around. But I wonder if the exclusivity of tracking data (both in terms of availability and required expertise) means more people will focus on apps and software rather than the data science.
There was a time when managers saying 'expected goals' was enough to make analytics twitter very happy. But I think Christian Purslow's video on the Grealish sale at Aston Villa was a watershed moment in 'analytics' being publicly acceptable as a reason for decision-making. And not all decisions are good ones.
Purslow himself wasn't really putting forward a data argument for the signings that Villa made to replace Grealish. But his "we couldn't replace Grealish's talents with one player so we bought three players instead" is oh so close.
At the same time, the structures of football are very much being contested. I don't just mean FIFA's battles for a biennial World Cup. Multi-club structures are increasing in number, mainly involving English clubs as a central hub (as Rory Smith of the New York Times has written about), even as far down the pyramid as Fleetwood Town.
It strikes me that the types of people who would be interested in multi-club ownership structures would also be interested in heavily investing in analytics. And it strikes me that football even vaguely resembling an MLB-type farm system isn't something that most (English, at least) football fans will take to well. It'd just take one owner of a multi-club empire, who'd disregarded traditions just a little too much, to extol the virtues of data for things to get uncomfortable.
And then there's the flip-side of the plucky player using analytics to get a better contract: the bosses using data to stamp down calls for a raise. This seems like something that might happen behind closed doors and not see the light of day, but perhaps it might come out in a briefing war.
Get your "Are we the bad guys" memes ready.
Sure, this is a 'more of the same' prediction. But with The Athletic keeping on going with their data stuff (we got a graph with a p-value in December), Jamie Carragher using xG timelines on Sky Sports, and covid and political polling driving data-use elsewhere, there's no slowing this down.
I won't turn this into a media analysis newsletter, but lemme just draw attention to the fact that The Athletic is more similar to Reach (the ownership group of football.london, Manchester Evening News, and others) than it is to their national paper The Mirror. And in the TV world, Sky wedded to football coverage in exactly the opposite way to BT (who have been rumoured to be selling off BT Sport all year). So it seems unlikely that anyone will rival these two outlets' data use.
But everyone everywhere will keep picking up data stuff. That's just the way the world is now.
It's been noted before in football, and other sports, that the 'demographic pool' of players is somewhat different to the demographic pool of the data analysts and scientists. There's no good reason why that should be the case. Given how close clubs are (supposed to be) to their local communities, that could be a useful avenue for helping address this balance.
The analytics sphere is also pretty largely cisgender male, and programmes specifically aimed at encouraging cis women, non-binary people, and trans people into football analytics would be good too.
Credit to Brandon Kent, who has run a couple of Measurables Office Hours programmes to connect professionals in the biz with people from underrepresented groups.
[Please get in touch with any programmes or groups which may be relevant here, and I will add to the online version of this piece]
Oh, and one final one. The Get Goalside analytics newsletter will keep going. Hopefully bigger and better than ever. But really, as long as it has your readership, it'll be doing just fine as it is.
Read more posts like this in your inbox
Subscribe to the newsletter