It’s sort of become an annual tradition at Get Goalside to write about how the analytics industry is converging around one specific point. First it was pressure data. Then it was hybrid data (a mix of event data and tracking data). Now, well, it’s not a type of data at all.
A brief bit of context first. This newsletter is written off the back of attending the Training Ground Guru and StatsBomb conferences, which tend to have shiny, exciting speakers and therefore produce thoughts about the shiny, exciting side of the industry. And sometimes, like in 2021 when I wrote about Stats Perform having pressure data, I’m a little ahead of the data company’s own product announcement (‘Opta Vision’ splashed pressure intensity in 2022, which then got written about in the hybrid data newsletter).
Better to be seven months early than late. Send me an email if you want an accidental peek into any other company’s roadmap.
So back to the newsletter: everyone’s a decision-maker now.
I don't say that just because StatsBomb conference’s keynote talk, from Zelus Analytics’ Luke Bornn, was ‘cognitive biases as Taylor Swift lyrics’. (Get Goalside would like to make clear its firm support of tenuous and extended metaphors). Let’s head over to three of the most recent data-centred consultancies to start operating in the sport, and see what they say about themselves in the opening line of their website copy. Bold emphasis is mine:
SRC FTBL, whose Ravi Ramineni also spoke (about decision-making) at the StatsBomb conference: “SRC is a football consultancy and data provider. Our mission is to use data and analytics to help clubs make optimal decisions.”
Ludonautics, the Michael Edwards and Ian Graham venture aiming to be the best spin-off from a Liverpool institution since Wings: “Ludonautics is a sports advisory business dedicated to helping sporting organisations improve their decision making capabilities through access to insightful statistical analysis.”
Prospect Sport, who recently hired former-RB Leipzig and Stats Perform data scientist Tom Worville as their Head of Football: “Prospect are revolutionising decision-making in sport with the help of artificial intelligence.”
I’ll know who to call next time I don’t know what to have for tea.
(Just so you don't think this is some kind of selection bias, Analytics FC, a much earlier entrant into the space, calls itself a "football analytics specialist with proven experience in delivering data-driven solutions". My own paymasters, Twenty3, keep it short and sweet with "We maximise the potential of sports data." Even Zelus Analytics care not about your decisions, they're just "building the world’s best sports intelligence platform.")
As with any convergence, there are multiple reasons behind it.
To a large extent, analytics has always been about making better decisions (only very occasionally about helping players execute better). And analysts have always talked about this too, going on about the path of data usage going from descriptive, to predictive, to prescriptive. First you can use data to describe what happened, then to predict what will happen, and finally to advise on what you should do in pursuit of a certain aim.
On top of that, there’s the other thing that analytics practitioners have always said: it’s no use having data if it doesn’t get listened to. That’s both an internal thing and an external thing, whether getting a manager to pay attention to your graphs or getting someone to buy your products. In either case, getting the buy-in of the people who hire, fire, and sign off budgets can save you a lot of hassle and wasted time. And I'd guess they're very receptive to phrases like 'decision-making'.
Finally, it's also quite simply a story about how established football analytics is. Everybody named so far in this article has close to (or more than) a decade’s experience around professional sports in a data or data-adjacent role. If we include Bornn’s research papers on NBA analytics while an assistant professor at Harvard (which we should, because one was about defending), all of their pro sports careers started while current Arsenal men's manager Mikel Arteta was still a professional footballer. These are people who've worked for a long time, in several cases been at teams who've won (multiple) major trophies, and therefore have the heft to go right to where decisions get made.
So, there’s all of that. Decision-making. Not ‘using data better’ or even applying specific sets of expertise like ‘helping sign the best players’. Just decision-making. "Drop the The; it's cleaner".
Football will probably be better for it. Not least because clubs often simultaneously need good decisions while being scared of anything but the usual ones. Lee Mooney of MUD Analytics (formerly of Manchester City) said, during a Training Ground Guru conference panel, that clubs are a bit like nuclear power stations: people are scared to tinker with them in case things go horribly wrong. The point of good decision-making, I suppose, is to work out how close the reactor actually is to blowing up, reduce any actual risk, and then improve the low-danger areas in the meantime.
(It may also be why English clubs that embrace data tend either not to be at the same risk of catastrophic nuclear collapse (Man City in one sense, Brentford and Brighton in a different one) or have already gone through the process of blowing up (Liverpool, whose data journey began after a debt crisis-induced ownership change in 2010; Wigan, who are currently working with Prospect, after their own ownership problems saw them docked points for not paying staff on time under the old regime; Arsenal, who bought StatDNA months after the third successive May date of St Totteringham’s Day, in 2012, and then hired the topically data-inclined Sven Mislintat months after the first season without a St Totteringham’s Day since 1995, in 2017; Manchester United, who announced the hiring of their Director of Data Science a few months after Manchester City reached their first European Cup final in 2021))
While ‘analytics as decisions’ is a fun way to understand the business landscape, it also made me think about something more fundamental. It's a framing which I think explains a lot about the course analytics has gone through so far, and may give a better idea of how it will progress from here.
Is ‘analytics’ tech, or is it an information resource?
It’s been easy to think of it as the former for most of the ‘boom’ since the mid-2010s, evolving in public consciousness alongside social media, grandparents getting smart phones, and AI. But while there’s tech involved, analytics obviously fits better alongside other ‘movements’ of information and ideas.
The internet, computers, television: none of these took off because people understood them. In fact, their most rapid growth happened only once they could provide value without people understanding how they work. There's a reason ordinary people don't know what Linux is.
Now think about ideas like cognitive biases, microaggressions, issues with capitalism: you have to understand them to get value from them. Although the general population might not have as sophisticated an understanding as experts, a big driver in them taking off in the last decade has been a mix of pop-sci books and Twitter. In other, less medium-specific terms: because enough people got exposed to enough of the information to understand it.
This framing is meaningful because if analytics was a technology, then simply giving users better interfaces would be the revolution. And is that what happens? No: people don’t get good at smartly applying data to football problems because they have a login to software with possession value models. They get good because they hire a decision-making consulta- no, wait. They get good because enough of the concepts get enough of an explanation that they can understand the information and apply it.
Does this matter - like actually matter - in the grand scheme of things? Not really. Only as much as you care about the application of football analytics beyond your ability to win more games or earn a living. But it does mean that “why aren’t more clubs using data well” isn’t going to be solved by better tech alone, or by the existence of a handful of smart people in the industry.
There's an elephant in the room here. The title of this newsletter is clearly false. Everyone is obviously not a decision-maker now.
Most people are, and will continue to be, far, far from the room where it happens, falling into one of three categories.
Two are simple: those who feed into a decision-making process earlier in the chain, and those who don’t. Analysts who create reports which get read by senior coaches will be in the first group; analysts who create reports which get a cursory glance at best will be in the second. We know where fans tend to find themselves. Not everyone can feed into every decision-making process, football is full of people giving miles to gain an inch or being told their input is listened to when it’s merely being recorded.
The third and final group are the facilitators. Data engineers, report-creation tools, data providers themselves. On this, the StatsBomb conference – being a branded event – had some news: that StatsBomb will be doing football tracking data for themselves in the near future. That means that after basic event data, 360 data, and stop-motion gridiron data, the company are filling up their football data Pokédex fast (someone told me afterwards in a shady corner of the pub that if you press up, down, left, left, right, up, down while StatsBomb IQ loads then you get Gaelic football stats too).
An amusing quirk of facilitators is that a rapid rise in their number may end up producing 'two steps forward, one step back' situations. New data sources need to be examined and studied for their peculiarities; new APIs may not be well supported; new products may not have all of the features that were demo-ed. To take a point from Ramineni’s talk, decisions compound, for better or worse. Rejigging your work processes to fit in a different product or service is likely to be one of those compounding choices.
How long will it be until the average club uses the same core set of data for, say, five years in a row? (Will that be a new marketing point: 'you know us, you have a track record of data with us, you know how to navigate our bugs'?). It may not even be within the club's control, depending on what league-wide deals are in place. And even then, even if a club is happy with their current provider(s) and able to stay with them, the aspects of that data they use every week will probably change. New models, new stats, new events.
A lot of new decisions to make.
If only there were someone who could help you out with those…
Thanks for reading, or scrolling right to the bottom. Get Goalside will be going on a hiatus until the spring, at least. It’s been a fun short run of newsletters with bringing facts to the added time discourse, riffing on Oreos and innovation, and playing with publicly-available tracking datasets in both Skillcorner and Metrica varieties. Stay well.
“The wedding was charming if a little gauche / There’s only so far new money goes,” –Taylor Swift, warning against assuming new wealthy club owners will solve all problems.
"What's hard about art is getting any good – and then getting any better. What's hard is solving problems with infinite solutions, and your finite brain." – Elisa Gabbert, 'Why Write'.
“I had not the hardihood to suggest to Dora’s father that possibly we might even improve the world a little, if we got up early in the morning, and took off our coats to the work.” – Charles Dickens, David Copperfield.