It was the Elite Minds in Sports conference this week at Arsenal’s Emirates Stadium and staff from several top clubs (Arsenal, Chelsea, Liverpool and Everton) were on hand to give their views on topics such as talent acquisition, injury management and analytics.
Importantly, also present was someone from mainstream media: Sean Ingle of the Guardian. Ingle looks to be increasingly interested in the area of football analytics, and follows a number of professionals in the industry and indeed a number of amateur bloggers on Twitter. This week in the Guardian online edition he went so far as to quote Statsbomb’s Ted Knutson in a piece on Crystal Palace. This is major progress over here. Over in Holland, Simon Gleave, Sander Ijtsma and Tijs Rokers already have regular columns in national daily De Volkskrant.
Anyway, James Smith, Everton’s Head of Tactical Scouting was first up at the conference and it was his comments tweeted by Ingle that caused a bit of a stir in the community. Here’s a taste or three:
“We’re not using sophisticated regression analysis when scouting players – that’s what bigger clubs do.”
“We do more GSCE maths stuff – averages, bar charts – when it comes to player analysis but that will change.”
“We don’t spend enough on staff in English football. We spend so much on wages, agents and transfer fees. It’s crazy. ”
Against that GCSE comment, Ingle hashtagged: #eliteminds. Whether that was done firmly tongue-in-cheek or not is unclear. But the comments certainly reverberated around the amateur blogging community. There were howls of derision at clubs’ stupidity to recognise the importance of or invest in analytics. The community had a point too. However, it led me to cringe in remembrance of an old memory I do my best to keep locked away – my first ‘proper interview’.
After finishing my degree in summer 1998, I made the decision to extend 3 years of bumming around by another month to watch the World Cup. On the Friday before the final, my friend chucked me an application form for a job he wasn’t interested in as he was off to law school. I filled it in, and sent it off. They rang back Monday offering me a short term contract without even speaking to me. Within a year I was a permanent member of staff and wanted to move on internally for a better job.
Well, in my first ‘proper’ interview, fresh and thoroughly wet behind the ears, I pretty much slagged off anything and everything about my current department and told them how it should have been done if only they were competent enough. Like the blogging community this week, I had good points and I was probably right. Funnily enough, though, I didn’t get the job. It’s worth bearing in mind.
Smith and his colleagues, Steve Brown, and Dan Hargreaves featured in Simon Kuper’s excellent piece in the FT and even in the opening chapters of Chris Anderson and David Sally’s The Numbers Game. Now, whichever way you look at it, it’s pretty clear that the methods described by Smith have been working for Everton. Everton haven’t done anything hugely sophisticated with numbers, yet have managed to consistently be ‘best of the rest’ once the big boys have shared the main spoils. Doing the traditional stuff well has given them an edge. The bloggers also seemed to ignore Smith’s words at the end of the GCSE quote – that things will change.
This requires buy in from the lead decision makers. After the conference Smith was quoted as saying: “Bloggers need to find metrics that are so compelling the critics can’t say no.” The bloggers are right in responding that there also needs to be at least a small leap of faith shown by decision makers. There’s a lot talked about secrecy and competitive edge in this industry so do clubs really want great stuff being published for everyone to see before they buy in?
The models we’ve made here at Differentgame are essentially a poor man’s version of what Prozone and Opta already do and what a club will have to generally pay extra for. There are regular questions about the nuance of SPAM as a simple model. What about headers compared to shots by foot for example? Colin Trainor and Constantinos Chappas over at Statsbomb go in to much further detail than I did for SPAM in their expected goal model – much smaller zones, distinguishing between headers and foot shots and shot placement etc. Our Chance Creation Model now factors in how the ball was delivered to the shooter. But the reality is, all these extra layers equal very small differences to outcomes – even over a full season. Basic models work.
There seems to be an insistence that clubs need staff with maths PhDs to look at hugely sophisticated data. It would help, but it’s nothing like essential in getting on the road to analytics. It just takes a decent idea, time, and hard work to look at the data available.
OptaPro are hosting their inaugral Analytics Forum in February next year. There’ll be impressive stuff on show for sure, but will there be anything so compelling that critics can’t say no? I have my doubts. The critics seem to want a magic number giving them the keys to the football universe. It simply doesn’t exist.
There’s plenty of compelling evidence pointing to the fact millions are wasted each year on buys that don’t work out. The critics in boardrooms up and down the country say an awful lot of ‘yes’ to that on a regular basis.
So here’s one basic pitch to the management and boardroom: “Invest another fifty to a hundred grand in analytics and save yourself a few million on transfers every season.”
Lot’s of buys don’t work out because a lot of what goes on the pitch is pretty random. There’s a lot of luck going on out there and it’s a game of chance a lot of the time. Analytics will help to load the dice slightly more in your favour.
Both the clubs and the blogging community might want to give each other more of a fighting chance too.