Goalkeeper statistics

[blaaklint]

http://www.sloansportsconference.com/wp-content/uploads/2019/02/Data-Driven-Goalkeeper-Evaluation-Framework-1.pdf

 

It's quite a long read but I thought that this might prove very interesting for anyone who has an interest in stats in football. It also says some quite interesting things about Schmeichel's ability in relation to other keepers in the PL.

[ealingfox]

Go on...

[Costock_Fox]

I have the attention span of a puppy on speed, can you summarise what it says?

 

Edit, I’ve just clicked on the link and there isn’t a hope in hell im reading that  

Edited 4 March 2019 by Costock_Fox

[one F in Fox]

30 minutes ago, blaaklint said:

http://www.sloansportsconference.com/wp-content/uploads/2019/02/Data-Driven-Goalkeeper-Evaluation-Framework-1.pdf

 

It's quite a long read but I thought that this might prove very interesting for anyone who has an interest in stats in football. It also says some quite interesting things about Schmeichel's ability in relation to other keepers in the PL.

Can you just cut to the chase and tell us either how good or crap Schmeichel is?

 

Thanks, ?

[Phube]

Christ that’s 43 pages of in depth stats and analysis!!! 

[tylesta]

**** reading that. He can't kick end thread

[Mike Oxlong]

That reads like a Thracian post

[Costock_Fox]

3 minutes ago, Mike Oxlong said:

That reads like a Thracian post

Without the homophobia (probably)

[Dr The Singh]

1 in 11 players are goal keepers in a team

[st albans fox]

4 minutes ago, Dr The Singh said:

1 in 11 players are goal keepers in a team

2 in 18 ........

[Sly]

In summary:

 

Kasper cost us 6.2 goals last season overall. By relative comparasion, Nick Pope saved Burnley, 11.5 goals, Jacob Butland 5.6. So we’d have been better off with one of them. 

 

De Gea is the best shot stopper in the league, however patrols in the 6 yard box and doesn’t really leave the area. Has average passing ability, Ederson is the best distributor but a below standard shot stopper. Adds massively to Man City’s Attack philosophy. 

 

 

 

 

[fuchsntf]

8 minutes ago, Sly said:

In summary:

 

Kasper cost us 6.2 goals last season overall. By relative comparasion, Nick Pope saved Burnley, 11.5 goals, Jacob Butland 5.6. So we’d have been better off with one of them. 

 

De Gea is the best shot stopper in the league, however patrols in the 6 yard box and doesn’t really leave the area. Has average passing ability, Ederson is the best distributor but a below standard shot stopper. Adds massively to Man City’s Attack philosophy. 

 

 

 

 

How many did Pope cost Burnley,how many did Schmeichel save us....!!!  Ffs!

[blaaklint]

All the data is taken from last season, but the most interesting are these three screenshots:

 

This first table is about shot-stopping ability. Essentially, every shot on target that isn't blocked is assigned a post-shot expected goal value (PSxG) - that is, the quality of that shot based on its trajectory, speed, how many players are in the way etc. The total PSxG is added together and the actual goals (GA) is subtracted. This produces a goals-saved above average (GSAA) for each goalkeeper, which is then divided by shots on target throughout the whole system (SOT) in order to standardise it to a percentage (GSAA%).

 

tl;dr - the higher the GSAA%, the better at shot-stopping a keeper is. Schmeichel is not very good; David de Gea and Nick Pope are gods.

 

This second table relates to ability at claiming crosses and commanding the area; it's also a bit simpler. The expected claims (xC) for a keeper are a little like the aforementioned (PSxG): they are the likelihood that each "claimable" (a collectable ball into the penalty: i.e. a cross or long ball) is collected, based on starting position of cross, trajectory, angle, pace and type, etc. The number of claims made is subtracted to give a claimables collected above average (CLCAA) value, which is then divided by the number of claimables to give a standardised claimables collected above average percentage (CLCAA%).

 

tl;dr - the higher the CLCAA%, the better the keeper is at collecting balls into the box. Schmeichel is above average but, for the most part, all keepers are much of a muchness.

 

This final graph is probably the most interesting, given last match's events. It charts the total positive outcome contribution (i.e. how likely a pass from a goalkeeper is to end having a positive outcome, defined as a free kick, a corner, a shot or a penalty) versus the change in pass length of a keeper when under pressure (i.e. does the goalkeeper go longer when under pressure or slightly shorter - does pressure make them behave differently to normal?).

 

tl;dr - moves involving Schmeichel typically end up pretty successfully. He is also probably the worst keeper with the ball at his feet when under pressure in the PL.

 

Note: Stats aren't perfect: some have arbitrary parameters (the weightings on a positive outcome for instance), others tell us things that we already know (David de Gea is a shot-stopping beast and Ben Foster is pretty much awful at everything). However, they do tell us some interesting and unexpected things which help to bring a greater, albeit flawed, idea of, in this case, how good a goalkeeper is. I think they also show just how much we are influenced by our own sporting preconceptions. For instance, I was particularly interested to see that Ederson did so badly shot-stopping-wise, but when you consider that Manchester City conceded so few decent opportunities last year, and with how good he is on the ball and at commanding his area (especially important given how high Man City played), it doesn't really matter to them. Schmeichel, it seems, although these are purely last year's stats, is an average keeper: relatively poor at shot-stopping but, most of the time, good on the ball and at commanding his area; however, a certain level of sentimentality does make me think that he is better than is shown above.

 

Are stats actually useful? Who knows. In the end, does anyone really care? I am, after all, a football fan rather than a statistics fan.

[Costock_Fox]

17 minutes ago, fuchsntf said:

How many did Pope cost Burnley,how many did Schmeichel save us....!!!  Ffs!

I presume he means an aggregate, pretty damning if so really.

[lgfualol]

Where is he in the shout at teammates after personal errors table. Surprised he didnt square up to Chilwell after slicing the ball and losing us the game.

[womp]

havent got through it all yet but yes it is interesting.

the ramifications of not having a decent keeper are bigger than you think for teams.

who would of thought you could justify De gea cost but they have.

 

[Deeg67]

Wish we’d gone after Fabianski when we had the chance.

[worthosoriginals]

44 minutes ago, Sly said:

In summary:

 

Kasper cost us 6.2 goals last season overall. By relative comparasion, Nick Pope saved Burnley, 11.5 goals, Jacob Butland 5.6. So we’d have been better off with one of them. 

 

De Gea is the best shot stopper in the league, however patrols in the 6 yard box and doesn’t really leave the area. Has average passing ability, Ederson is the best distributor but a below standard shot stopper. Adds massively to Man City’s Attack philosophy. 

 

 

 

 

is this per game?

[ThaiFox]

Statistics.......

 

They can say whatever they want you to say.

 

Number of shots saved? What about the quality of shots, how powerful, on the strikers 'wrong' foot etc?

 

Number of goals let in? What about defenders mistakes? It stands to reason a team with outstanding defenders has a better goalkeeper!

 

Number of great saves? Depends on a number of factors, including state of the match, the pitch, luck, skill, etc, etc.

 

Distribution? Depends on what you do, pass it 2 yards, or a 100 yard boot up the field, or if it directly leads to a goal.

 

No-one remembers these things, but they remember the result. Could you really care less we had 27 shots in a match but lost 1-0?

 

It's just like possession. It means absolutely nothing.

 

All that matters in the match end scoreline. 

 

 

 

 

Edited 5 March 2019 by ThaiFox

[Sly]

6 hours ago, fuchsntf said:

How many did Pope cost Burnley,how many did Schmeichel save us....!!!  Ffs!

 

6 hours ago, worthosoriginals said:

is this per game?

It’s an over the season average. 

 

So Pope was worth 17.7 more goals saved over Kasper. That’s pretty damming. 

 

Kasper is also woeful, if not the worst in the league under pressure with the ball at his feet.

 

It’s amazing what the money ball approach shows you. 

 

Something worth mentioning however, Mark Warbuton uses this type of analysis lot. It hasn’t  really worked for him long term. 

 

Will Brendan do a Pep and bun him off, ala Pep / Hart.

[Guest]

1 hour ago, ThaiFox said:

Statistics.......

 

They can say whatever they want you to say.

 

Number of shots saved? What about the quality of shots, how powerful, on the strikers 'wrong' foot etc?

 

Number of goals let in? What about defenders mistakes? It stands to reason a team with outstanding defenders has a better goalkeeper!

 

Number of great saves? Depends on a number of factors, including state of the match, the pitch, luck, skill, etc, etc.

 

Distribution? Depends on what you do, pass it 2 yards, or a 100 yard boot up the field, or if it directly leads to a goal.

 

No-one remembers these things, but they remember the result. Could you really care less we had 27 shots in a match but lost 1-0?

 

It's just like possession. It means absolutely nothing.

 

All that matters in the match end scoreline. 

 

 

 

 

Are you Kasper's Dad?

[blaaklint]

1 hour ago, ThaiFox said:

Statistics.......

 

They can say whatever they want you to say.

 

Number of shots saved? What about the quality of shots, how powerful, on the strikers 'wrong' foot etc?

 

Number of goals let in? What about defenders mistakes? It stands to reason a team with outstanding defenders has a better goalkeeper!

 

Number of great saves? Depends on a number of factors, including state of the match, the pitch, luck, skill, etc, etc.

 

Distribution? Depends on what you do, pass it 2 yards, or a 100 yard boot up the field, or if it directly leads to a goal.

 

No-one remembers these things, but they remember the result. Could you really care less we had 27 shots in a match but lost 1-0?

 

It's just like possession. It means absolutely nothing.

 

All that matters in the match end scoreline. 

 

 

 

 

But the entire point of the use of PSxG is that it controls for position taken from, how many defenders are in the way, where the ball would have entered the goal, pace and so on.

 

A defender’s mistake makes no difference, since it’s the quality of the actual shot itself being measured. If a defender loses the ball and the attacker skies it, that says mothing about the keeper. If they take a shot down the centre of the goal from a long way with no pace, and a goalkeeper doen’t save it, the goalkeeper has underperformed, or, if the shot’s a screamer and is saved, the goalkeeper has overperformed.

 

Now, the most important number is indeed the score. But stats are interesting, if only because in the long run they even out and, in this case, attempt to quantitatively measure goalkeeper performance. The point is that, with a different keeper, some better shots might be saved and some worse shots let in.

[RumbleFox]

Some people really don't understand statistics. X 

[Dahnsouff]

9 hours ago, blaaklint said:

http://www.sloansportsconference.com/wp-content/uploads/2019/02/Data-Driven-Goalkeeper-Evaluation-Framework-1.pdf

 

It's quite a long read but I thought that this might prove very interesting for anyone who has an interest in stats in football. It also says some quite interesting things about Schmeichel's ability in relation to other keepers in the PL.

Fibber. That’s not about goalkeeper statistics at all.

its a scientific study to gauge the reading efforts made by the Twitter generation

[RumbleFox]

9 hours ago, blaaklint said:

All the data is taken from last season, but the most interesting are these three screenshots:

 

This first table is about shot-stopping ability. Essentially, every shot on target that isn't blocked is assigned a post-shot expected goal value (PSxG) - that is, the quality of that shot based on its trajectory, speed, how many players are in the way etc. The total PSxG is added together and the actual goals (GA) is subtracted. This produces a goals-saved above average (GSAA) for each goalkeeper, which is then divided by shots on target throughout the whole system (SOT) in order to standardise it to a percentage (GSAA%).

 

tl;dr - the higher the GSAA%, the better at shot-stopping a keeper is. Schmeichel is not very good; David de Gea and Nick Pope are gods.

 

This second table relates to ability at claiming crosses and commanding the area; it's also a bit simpler. The expected claims (xC) for a keeper are a little like the aforementioned (PSxG): they are the likelihood that each "claimable" (a collectable ball into the penalty: i.e. a cross or long ball) is collected, based on starting position of cross, trajectory, angle, pace and type, etc. The number of claims made is subtracted to give a claimables collected above average (CLCAA) value, which is then divided by the number of claimables to give a standardised claimables collected above average percentage (CLCAA%).

 

tl;dr - the higher the CLCAA%, the better the keeper is at collecting balls into the box. Schmeichel is above average but, for the most part, all keepers are much of a muchness.

 

This final graph is probably the most interesting, given last match's events. It charts the total positive outcome contribution (i.e. how likely a pass from a goalkeeper is to end having a positive outcome, defined as a free kick, a corner, a shot or a penalty) versus the change in pass length of a keeper when under pressure (i.e. does the goalkeeper go longer when under pressure or slightly shorter - does pressure make them behave differently to normal?).

 

tl;dr - moves involving Schmeichel typically end up pretty successfully. He is also probably the worst keeper with the ball at his feet when under pressure in the PL.

 

Note: Stats aren't perfect: some have arbitrary parameters (the weightings on a positive outcome for instance), others tell us things that we already know (David de Gea is a shot-stopping beast and Ben Foster is pretty much awful at everything). However, they do tell us some interesting and unexpected things which help to bring a greater, albeit flawed, idea of, in this case, how good a goalkeeper is. I think they also show just how much we are influenced by our own sporting preconceptions. For instance, I was particularly interested to see that Ederson did so badly shot-stopping-wise, but when you consider that Manchester City conceded so few decent opportunities last year, and with how good he is on the ball and at commanding his area (especially important given how high Man City played), it doesn't really matter to them. Schmeichel, it seems, although these are purely last year's stats, is an average keeper: relatively poor at shot-stopping but, most of the time, good on the ball and at commanding his area; however, a certain level of sentimentality does make me think that he is better than is shown above.

 

Are stats actually useful? Who knows. In the end, does anyone really care? I am, after all, a football fan rather than a statistics fan.

Top post, well done. X