Data in football

[swan65split]

Data....unpredictable......as the data of future opponents etc is not known, and is always changeable.

only one thing is certain, ...........

 

[J_B]

Liverpool has let the way on this for over a decade. Brighton and Brentford have been very successful more recently with this approach.
Those clubs employ the right people who understand it's merit and limitations. Without the skills and understanding of what it can and can't do it's largely useless.

 

[Professor]

It may work for American sports like baseball, where are lot of the KPIs can be reduced to decisions like "does he hit the ball more often than not? Y/N". But as we all know, football is full of little quirks and oddities, many of which you can't quantify.

Even the stats you can quantify often don't tell you the full picture. If you had a look at Grimes on a spreadsheet, you'd probably get told here's a guy who never gives the ball away and is always fit. What the data won't tell you is that he rarely loses the ball because he plays with zero risk and passes backwards and sideways most of the time, and he's always fit because he never engages physically. He's a pretty ineffectual player - but some of the stats might tell you otherwise.

There is no substitute for the good old fashioned eyeball test. Not sure there's enough of that going on at our club anymore, and not for a long time.

Which is still data but qualitative rather than quantitative. It's harder to model which is why it's less fashionable. Just using quantitative data often leads to mistakes. McDonalds is the largest restaurant chain in the world by number of outlets and sales. Using those datasets may suggest it's the best restaurant in the world. It's clearly not (the current holder in Disfrutar in Barcelona) and there are better places in most towns. So having someone with understanding and expertise is always needed.

 

[JackSomething]

Liverpool has let the way on this for over a decade. Brighton and Brentford have been very successful more recently with this approach.
Those clubs employ the right people who understand it's merit and limitations. Without the skills and understanding of what it can and can't do it's largely useless.

Exactly. It's not the data that is the problem, it's using it properly.

Brighton (as you point out) use it to identify lesser known young players, who they sign at a relatively cheap price with the aim of them becoming first team players and valuable assets. It's worked brilliantly with the likes of Caceido, Mitouma and many more.

Meanwhile, Man United take that same data, analyse over 800 right backs and decide Aaron Wan-Bissaka is both the right option and worth £50m.

 

[Ringwood]

At least the way we play would kill AI

Can you imagine AI trying to replicate how we play

 

[Gors Buster]

Montague spoke about the weighting of data.
Gaultier Ott’s data may have been off the chart for the Portuguese 2nd division but what does that mean? Looking at YouTube videos, they look like Welsh League grounds.
If he was that good then surely someone in the Portuguese first division would have signed him.

 

[exiledclaseboy]

Liverpool has let the way on this for over a decade. Brighton and Brentford have been very successful more recently with this approach.
Those clubs employ the right people who understand its merit and limitations. Without the skills and understanding of what it can and can't do it's largely useless.

You’ve hit the crux there. Understanding the limitations is equally as important as understanding the merits.

 

[K23]

It’s a broad and deep subject, without a doubt. My professional life crosses over F1 telemetry and that is mind blowing in terms of the collection methods, processing methods and the ultra fine marginal gains that it delivers. Whilst football and motor sports are worlds apart, but that’s not to say that there is no place in football for a data driven strategy, however it’s not the only driver.

It may not be the only driver but if it's cheaper than flying people to the continent to watch a particular player a few times, then it's probably the only driver a club like ours will use.

 

[TheLoneRanger]

Going back to 2015, Arsene Wenger had a team of 20 people working on data analysis @ Arsenal.

 

[Whaleoilbeefhooked]

Why such a big push towards it?

I see lots of things going wrong with the use of data.
Who puts the data in and is their judgement right in the first instance.
Building a team around data, you are relying on a person to control the style that isn't necessarily correct in the first place to then pick players from a database to fit it.

I understand it's an handy tool to have but it seems concerning that so many people seem to be relying heavily on it, especially given alot of trust needs to be put in people entering the correct information in the first place, a process I feel can produce many mistakes and surely many areas would fall on the data inputers opinion.

The old adage of shit in shit out, still stands