Do you think stats ruin the game?

BoroMart said:
exactly that, making decisions on one metric is a poor thing to do, as you lose all context.

Charles Hughes in the late 70s got data from 1,000s of matches that showed that the majority of goals are scored from 5 or less passes. He pushed the FA to use the mantra of "get it forward fast", direct football, this culminated in Graham ' hit it long' Taylor becoming England manager.

The problem is that years later, proper statisticians looked at the data, and analysis showed that there are many many times more '5 or less pass sequences' in a match than say '20 pass sequences'. When you deep dive into the data the chances of "20 passes leading to a goal" was actually many, many times higher than "5 or less passes leading to a goal".

He'd ben a right Charley and used a single metric to define the philosophy of English football for a decade, they even taught the up and coming coaches like Warnock that this was the best chance of success, Charles set english football back 20 years.
Click to expand...
I think I read about that in Inverting the Pyramid (really good book on football tactics). The thing is, that philosophy was (and still is) right for clubs who would be on the back foot in games with average footballers. But it does nothing for you when you have a side full of quality players who are going to be in control of a game against a side who will drop back and try to frustrate you.

Context is key in all of these things. Signing a tall striker who has scored 20 goals in a season where he had an XG of 15 is great when you have two wingers with a high xA. Not so great when you've already got a striker scoring above his XG with a midfield who averages around a 3 xA each over a season. Even less great if you sign him to play as a lone striker when he currently plays in a front two with his other striker creating a lot of space for him to get into.

The stats today are far better than they were in the past. There's just so much data collected. And most clubs do have data analysts who understand how to avoid misinterpreting the data. The problem is, I'm not sure how many data analysts out there have good footballing brains. I work as a freelancer in AI/Machine Learning and I always tell my clients they need to find someone to interpret the data who understands the industry. I'll usually turn down the job of interpreting the data if I don't know enough about the industry. Sometimes it takes someone with industry knowledge to point out where things that are theoretically statistically significant aren't actually significant I the real world.
 
rbmfc1985 said:
I think I read about that in Inverting the Pyramid (really good book on football tactics). The thing is, that philosophy was (and still is) right for clubs who would be on the back foot in games with average footballers. But it does nothing for you when you have a side full of quality players who are going to be in control of a game against a side who will drop back and try to frustrate you.

Context is key in all of these things. Signing a tall striker who has scored 20 goals in a season where he had an XG of 15 is great when you have two wingers with a high xA. Not so great when you've already got a striker scoring above his XG with a midfield who averages around a 3 xA each over a season. Even less great if you sign him to play as a lone striker when he currently plays in a front two with his other striker creating a lot of space for him to get into.

The stats today are far better than they were in the past. There's just so much data collected. And most clubs do have data analysts who understand how to avoid misinterpreting the data. The problem is, I'm not sure how many data analysts out there have good footballing brains. I work as a freelancer in AI/Machine Learning and I always tell my clients they need to find someone to interpret the data who understands the industry. I'll usually turn down the job of interpreting the data if I don't know enough about the industry. Sometimes it takes someone with industry knowledge to point out where things that are theoretically statistically significant aren't actually significant I the real world.
Click to expand...
Yes Inverting the pyramid is a good book, The Mixer is pretty good as well if you haven't read that one.

Agreed on context and industry understanding, and that comes with experience. The football industry is rapidly professionalising, but there's probably quite a small base of experienced, specialist football data analysts still and will be for another 10 years.

People don't realise that top clubs even record and analyse every training session, when a manager says player X isn't showing enough in training, he usually has the data to back it up with heat maps, distance covered, top speed, average speed, plus highlights clips.
 
BoroMart said:
Yes Inverting the pyramid is a good book, The Mixer is pretty good as well if you haven't read that one.

Agreed on context and industry understanding, and that comes with experience. The football industry is rapidly professionalising, but there's probably quite a small base of experienced, specialist football data analysts still and will be for another 10 years.

People don't realise that top clubs even record and analyse every training session, when a manager says player X isn't showing enough in training, he usually has the data to back it up with heat maps, distance covered, top speed, average speed, plus highlights clips.
Click to expand...
Yeah, there's specialist software that can do a lot with recorded training sessions and matches. It can collect a lot of information that the data analysts can use. I'm assuming with what you've said already, that you're probably aware of that, maybe even more than I am. But a lot of people aren't aware how advanced most clubs are now with the data they collect on their players.

I've never read the Mixer, I'll have to check it out.
 
rbmfc1985 said:
Yeah, there's specialist software that can do a lot with recorded training sessions and matches
Click to expand...
Yes I've had a browse through the capabilities of Veo, it's something I'd like to do at some point, I'm looking at doing my FA Talent Identification courses over the next couple of years. Then do some Veo training and have a weekend/evening side job in that with the aim to have it in my pocket when I retire from my day job in 10 years.
 
As someone who believes in the power of data across every aspect of life, providing it is analysed, understood and utilised properly, you can probably guess my answer.

I do think football data is too often viewed in isolation however without looking for correlation. Any activity with a clear objective, such as football (scoring and stopping goals) can (and should be) reduced to a statistical analysis in my view. Teams who understand this tend to be those making the most progress at the extent of those who don't.

Of course you may think this leads to certain styles and patterns of play (it certainly can do) but important to remember that "entertainment" is not an objective of the sport.
 
Laughing said:
It's the number of goals a team would expect to score given the chances they created in a game. Easy chances have a high xg. 1 would be a certain goal and should never happen 0 is no chance of scoring and, probably shouldn't happen either,unless it's an own goal, I suppose with no opposition player involved.

If a team has 3 chances all with an xg of 0.5 it means 50% of the time you would expect a player to score that chance. The teams xg for the game would be 1.5,even if they scored 0 or 4 goals.
Click to expand...
That sounds very complicated to me- Who decides what an "easy" chance is. That sounds very subjective- or am I missing something.
 
Trug said:
That sounds very complicated to me- Who decides what an "easy" chance is. That sounds very subjective- or am I missing something.
Click to expand...
It’s subjective to n that someone made a subjective system, once that system is in place comparison between chances isn’t subjective because everything is measured to the same standards. So you can question the accuracy as a whole more than the subjective ness of the observer, if that makes sense
 
BoroMart said:
It’s subjective to n that someone made a subjective system, once that system is in place comparison between chances isn’t subjective because everything is measured to the same standards. So you can question the accuracy as a whole more than the subjective ness of the observer, if that makes sense
Click to expand...
I believe ai is now used to get consistency across xg. It used to be human watchers.

Still a poor stat though.
 
rbmfc1985 said:
That's why they use the xA stat as well. Clubs won't be making all of their decisions on goals and assists. They'll be looking at the xG and xA to try and get a striker who scores more than his xG and midfielders/wingers who have high xA stats.
Click to expand...
Is there a xT stat for tackles?
xF for fouls committed/fouled against? xS for saves by the keeper. We could go on.
I'm not knocking stats although they are not for me. I do realise that for people who have analytical minds they are useful information. They seem to be measuring the performance of human beings in the same way I used to measure the performance of machinery when I used to work.
 
Trug said:
Is there a xT stat for tackles?
xF for fouls committed/fouled against? xS for saves by the keeper. We could go on.
I'm not knocking stats although they are not for me. I do realise that for people who have analytical minds they are useful information. They seem to be measuring the performance of human beings in the same way I used to measure the performance of machinery when I used to work.
Click to expand...
Thats pretty much exactly what it does. It allows you to asign a number to an aspect of the game. They are great for anyone doing analytical analysis from a betting perspective.

Once you have numbers you can compare a with b directly.
 
You must log in or register to reply here.
Back
Top