Expected Goals

[LookMumImOnMCF.net]

Think this might be the article: http://www.bbc.co.uk/sport/football/40699431

I quite like that it tries to add a qualitative element , its not perfect, but then all stats can be rendered useless. Shots off target for example - were they decent shots or all from 35 yards?

I don't think the name "Expected Goals" is great, should be Chance Creation, or something better than that.

the arsenal game was a classic example of where expected goals falls short
we created 7 or 8 very good chances but only a couple of these led to a shot on goal

It does what it does though, you can only really argue against the methodology. The xG for that game was 2.2? Dunno what your argument is against it here.

As I say it's not perfect but it's an interesting debate to have. Stelling & co's dismissal of it earlier in the week is symptomatic of the dinosaur analysis we've put up with for too long

[nottsblue]

What's the xg for a penalty?

[LookMumImOnMCF.net]

What's the xg for a penalty?

0.76

[nottsblue]

What's the xg for a penalty?

0.76

At the risk of showing my ignorance, but does that imply that, on average, one out of every penalties is missed?

[LookMumImOnMCF.net]

What's the xg for a penalty?

0.76

At the risk of showing my ignorance, but does that imply that, on average, one out of every penalties is missed?

Not sure if you missed a "4" out of your post but if so then yeah.

Remember it's just a dataset of (I think I read) 5,000 pens. It doesn't take into a account an individual player's record, each of their xG will be different.

[nottsblue]

What's the xg for a penalty?

0.76

At the risk of showing my ignorance, but does that imply that, on average, one out of every penalties is missed?

Not sure if you missed a "4" out of your post but if so then yeah.

Remember it's just a dataset of (I think I read) 5,000 pens. It doesn't take into a account an individual player's record, each of their xG will be different.

Yes, did miss out the four. Interesting that an individual will have a different xg for a penalty. So by that token Yaya will be 1 xg on a penalty?

Conversely, does a goalkeepers success rate come into the equation or is it solely down to the attacking side?

[LookMumImOnMCF.net]

If he takes 10 pens and scores 10 then yeah, his xG would be 1. You could say 10 isn't enough data to go on really. 0.76 xG will be an average, most players will be better/ worse to some degree.

Individual stats like an individual player's skill or what foot they are for example, keeper in goal, isn't taken into account. The guy who came up with xG has admitted this is a drawback of xG.

Again I'm not saying it's perfect but it's a decent start ok imo

[PrezIke]

https://www.theguardian.com/football/bl ... SApp_Other

Interestingly enough, an article in The Guardian today on xG from “Real football men vs the nerdy analytics crowd … ”

The basic concept of expected goals is simple. It answers the question: should a player have scored that chance? Or, when you add all the chances together, should a team have won that match?...

The question of whether this stat is useful to you depends on what you want from a game. A lot of people, increasingly known as “real football people”, don’t need a stat to tell them what they can see with their own eyes. Others, the analytics crowd or “football nerds”, think it provides context. Especially if you didn’t actually watch the game.

Part if a comment posted I100% agree with:

I think it's quite funny that xG is derided so much by the so called "real football people", as it's the stat that actually probably best correlates with what you actually see with your eyes when watching.

I've done a fair bit of work on xG this year with regards to the Premiership games, and I can confirm, it ain't perfect. But then nothing is. And nothing ever will be with football, it's far too complex a game. But it ain't half bad, and the more complex the models that compute the xG, the more accurate it will become. As I said, it ain't half bad, it's predicting dips in results based on performance fairly consistently. And the opposite.

[patrickblue]

I think they should have an expected excuses stat.
Whinger would get top marks every game.

[patrickblue]

But then Moaning Mo will probably score pretty highly tonight.

[Foreverinbluedreams]

Think this might be the article: http://www.bbc.co.uk/sport/football/40699431

I quite like that it tries to add a qualitative element , its not perfect, but then all stats can be rendered useless. Shots off target for example - were they decent shots or all from 35 yards?

I don't think the name "Expected Goals" is great, should be Chance Creation, or something better than that.

the arsenal game was a classic example of where expected goals falls short
we created 7 or 8 very good chances but only a couple of these led to a shot on goal

It does what it does though, you can only really argue against the methodology. The xG for that game was 2.2? Dunno what your argument is against it here.

As I say it's not perfect but it's an interesting debate to have. Stelling & co's dismissal of it earlier in the week is symptomatic of the dinosaur analysis we've put up with for too long

The argument is that Sterling's chance when he had a tap in to an open goal would not factor in this analysis because no shot was taken, this is where it falls short.

As is acknowledged in that Guardian article, individual player's ability needs to be acknowledged for it really to be a useful analytic tool. For example if DeBruyne has a chance on the edge of the box the expectancy that he will score is far greater than if say Silva had it.

[Beefymcfc]

Didn't we invent this as part of our visionary football?

[Mikhail Chigorin]

Didn't we invent this as part of our visionary football?

Sort of 'holistic hypotheticals' ??

[Beefymcfc]

Didn't we invent this as part of our visionary football?

Sort of 'holistic hypotheticals' ??

That made me laugh.

But, I'm sure our club started pulling together certain stats which included these, which was a transformation in footballing analysis.

Could be wrong though, I usually am

[LookMumImOnMCF.net]

Think this might be the article: http://www.bbc.co.uk/sport/football/40699431

I quite like that it tries to add a qualitative element , its not perfect, but then all stats can be rendered useless. Shots off target for example - were they decent shots or all from 35 yards?

I don't think the name "Expected Goals" is great, should be Chance Creation, or something better than that.

the arsenal game was a classic example of where expected goals falls short
we created 7 or 8 very good chances but only a couple of these led to a shot on goal

It does what it does though, you can only really argue against the methodology. The xG for that game was 2.2? Dunno what your argument is against it here.

As I say it's not perfect but it's an interesting debate to have. Stelling & co's dismissal of it earlier in the week is symptomatic of the dinosaur analysis we've put up with for too long

The argument is that Sterling's chance when he had a tap in to an open goal would not factor in this analysis because no shot was taken, this is where it falls short.

As is acknowledged in that Guardian article, individual player's ability needs to be acknowledged for it really to be a useful analytic tool. For example if DeBruyne has a chance on the edge of the box the expectancy that he will score is far greater than if say Silva had it.

100% agree. But if a player air-kicks from two years out it's not registered as a shot on or off target either, this isn't just an xG anomaly, it's an issue for a lot of stats.

Likewise, individual ability isn't taken into account on the quality of the corner taker, free kick taker, shots on/ off target player etc etc. Ultimately I don't think any stat will tell the full story, just offer a bit more insight.

Real football men vs the nerdy analytics crowd …

MCF.net vs LookMum round 2,149,092

[Foreverinbluedreams]

"A lot of people, increasingly known as “real football people”, don’t need a stat to tell them what they can see with their own eyes. Others, the analytics crowd or “football nerds”, think it provides context. Especially if you didn’t actually watch the game."

Do I have to pick a side, can't I judge from what I've seen but acknowledge that stats can be useful too?

It does provide some context but that last line is just bollox, if you didn't watch the game then you can't analyse it with any degree of competence regardless of what stats you use.

[Justified logic]

Think this might be the article: http://www.bbc.co.uk/sport/football/40699431

I quite like that it tries to add a qualitative element , its not perfect, but then all stats can be rendered useless. Shots off target for example - were they decent shots or all from 35 yards?

I don't think the name "Expected Goals" is great, should be Chance Creation, or something better than that.

the arsenal game was a classic example of where expected goals falls short
we created 7 or 8 very good chances but only a couple of these led to a shot on goal

It does what it does though, you can only really argue against the methodology. The xG for that game was 2.2? Dunno what your argument is against it here.

As I say it's not perfect but it's an interesting debate to have. Stelling & co's dismissal of it earlier in the week is symptomatic of the dinosaur analysis we've put up with for too long

The argument is that Sterling's chance when he had a tap in to an open goal would not factor in this analysis because no shot was taken, this is where it falls short.

As is acknowledged in that Guardian article, individual player's ability needs to be acknowledged for it really to be a useful analytic tool. For example if DeBruyne has a chance on the edge of the box the expectancy that he will score is far greater than if say Silva had it.

100% agree. But if a player air-kicks from two years out it's not registered as a shot on or off target either, this isn't just an xG anomaly, it's an issue for a lot of stats.

Likewise, individual ability isn't taken into account on the quality of the corner taker, free kick taker, shots on/ off target player etc etc. Ultimately I don't think any stat will tell the full story, just offer a bit more insight.

We need an expected assist (xA) stat as well then.

[mr_nool]

"A lot of people, increasingly known as “real football people”, don’t need a stat to tell them what they can see with their own eyes. Others, the analytics crowd or “football nerds”, think it provides context. Especially if you didn’t actually watch the game."

Do I have to pick a side, can't I judge from what I've seen but acknowledge that stats can be useful too?

It does provide some context but that last line is just bollox, if you didn't watch the game then you can't analyse it with any degree of competence regardless of what stats you use.

Pretty Boy Lee can.

[iwasthere2012]

"A lot of people, increasingly known as “real football people”, don’t need a stat to tell them what they can see with their own eyes. Others, the analytics crowd or “football nerds”, think it provides context. Especially if you didn’t actually watch the game."

Do I have to pick a side, can't I judge from what I've seen but acknowledge that stats can be useful too?

It does provide some context but that last line is just bollox, if you didn't watch the game then you can't analyse it with any degree of competence regardless of what stats you use.

Pretty Boy Lee can.

Perhaps he can.

I’d go one further and say being at the game and watching the broader scale of things, enables you to form a much more accurate assessment of what is actually happening than seeing it on a tv screen which shows a much narrower view of the game.
I believe this, even despite the replays and different angles and slow motions etc.

I’m with FIBD in so far as a combination of statistical analysis and what a coach actually sees in training and in a match, with the benefit of video footage, is no doubt of high sports science value.

For the ordinary fan though, I sometimes despair of all the armchair analysis, quoting all sorts of stats as proof.

If you have any kind of football brain at all and you watch enough of the team to form more than a short term opinion, then I think there is enough evidence in front of your eyes week in week out. Stats help but I rarely use them as a sole prop for my opinions.
You can’t beat your own intuition or gut feel.

[Dameerto]

Didn't we invent this as part of our visionary football?

Sort of 'holistic hypotheticals' ??

That made me laugh.

But, I'm sure our club started pulling together certain stats which included these, which was a transformation in footballing analysis.

Could be wrong though, I usually am

A few seasons back the club released an enitre season's worth of data for people to experiement with from an analytical point of view - I have no idea if the guy who developed xG used that data though.