A Look at Goalkeeper Options and Replacing De Gea

[sifi36]

I've been analysing the goalkeeper data available through FBRef as I think goalkeeping is the easiest of all positions to analyse statistically.

I've taken some of the provided ones and come up with a few transformations to create a list of metrics to score goalkeepers from the top five leagues on their league performances over this season and the previous two to create reasonable size samples.

Metric	Definition	Limitations
Stopped Crosses %	Crosses Stopped divided by Crosses Faced	Doesn't take into account how "stoppable" a cross is, or failed cross claims
Distance Swept /90	Number of sweeping actions multiplied by average sweeping distance, divided by 90s played	doesn’t take into account failed sweeps which usually result in high scores quality chances
Shot Stopping	Post-shot xG faced minus goals against, divided by post-shot xG faced	FBRef post-shot xG model doesn't take into account shot location or power
Bad Touches /90	Miscontrols plus dispossessions per 90 played
Bad Touch Ratio	Miscontrols plus dispossessions divided by total touches
Outside Area Touches	Touches Outside penalty area divided by total touches
Outside Area Touches /90	Touches Outside penalty area per 90 played
Short and Medium Passes /90		Somewhat team style dependent
Short and Medium Share	A ratio of short and medium passes relative to total passes	Somewhat team style dependent
Short Accuracy	Completion percentage of short passes
Medium Accuracy	Completion percentage of medium passes
Long Accuracy	Completion percentage of long passes
Passing Tendancies	An average of the percentile rank of "Short and Medium Passes /90" and "Short and Medium Share"
Passing Accuracy	An average of the percentile rank of "Short Accuracy", "Medium Accuracy" and "Long Accuracy"
On Ball Comfort	An average of the percentile rank of "Bad Touches /90", "Bad Touch Ratio", "Outside Area Touches" and "Outside Area Touches /90"
Crosses	Percentile rank of "Stopped Crosses %"
Sweeping	Percentile rank of "Distance Swept /90"
Shot Stopping	Percentile rank of "Shot Stopping"
OVERALL SCORE UNWEIGHTED	An Average of each of the six above
OVERALL SCORE WEIGHTED	As above but each category is weighted by the amount specified

That then gives us the below table for the top 50 overall (Costa is highlighted as he doesn't play in a top five league and thus his data may or may not be comparable):

Lots of names you would expect, but also quite a few that I hadn't heard of: Casteels (contract expires 2024 and valued at 8m Euros on TransferMarkt), Samba (10m Euros) and Diouf (5m Euros) look very well balanced, with none of the six metrics being below average. Diogo Costa, Pau Lopez and to a lesser extent David Raya also look like good options for United.

David De Gea - 78th out of 90 goalkeepers that have played at least 50 games' worth of minutes:

If you're more of a traditionalist and value shot-stopping and command of area uber-alles, quadrupling the weight of those two metrics gives the following top 50

For a visual comparison of some we've been linked to and others we should be considering:

Conclusions

• David De Gea is either average or poor at everything a goalkeeper is required to do
• Upgrading on him can be done in an inexpensive manner
• Obviously scouting/eye-test and personality assessment would also be required before signing any of the above

You can download the Excel file from the link below and add your own weightings for each category in the first sheet ("Scoreboard"), I'd recommend filtering "90s" to at least 30 if not 50 to ensure the goalkeepers being looked at have reasonable sample sizes. The second ("Comparison Charts") allows you to create your own radar comparison chart for up to five of the goalkeepers included in the analysis.

Link to File

 

[Raw]

This is excellent stuff mate. I do hope we have people at the club doing something like this.

What is absolutely clear is that we need to upgrade De Gea.

 

[Zlatattack]

WOW! DDG has been that bad over the last 3 seasons!

From the table Costa looks like a no brainer.

 

[Meep]

Amazing that de Gea got the golden glove again this season with these stats. It is like he isn't the main reason for him getting new gloves. Costa or Diouf for me.

 

[sifi36]

This is excellent stuff mate. I do hope we have people at the club doing something like this.

What is absolutely clear is that we need to upgrade De Gea.

They have access to way more detailed data, I'd expect a lot more from a multi-billion pound enterprise, and yes, replacing De Gea is a must, it won't be expensive and there are loads of options (unlike striker).

WOW! DDG has been that bad over the last 3 seasons!

From the table Costa looks like a no brainer.

Agreed.

One more image, filtered for the top 50 goalkeepers that have played at least 50 games' worth of minutes.

 

[zaafi]

It's been years ago since De Gea was good. Having great reflexes doesn't make him a great keeper as some seem to believe. I've seen Pickford make world class saves again and again.

 

[JB7]

I always thought it was completely insane that we just watched Rulli go to Ajax in January for £8m without making any kind of a move, the data confirms it.

It also confirms that we don't have to spend big to considerably improve on the goalkeeper, which I've been trying to tell people for months.

Thank you for putting it in a clear, readable, easy format. Great work.

EDIT - One point on Raya as you mentioned him, he scores pretty well despite two very low scores in terms of the two passing options, those two scores are realistically products of the style of Brentford's play. His game is either to hit the target man (Toney) or hit space behind the full backs - which Ramsdale does a lot of - so half the time he isn't even aiming to keep possession, it's to turn the full backs and for the attackers to win the ball high up the pitch. In a team like United, those metrics would increase significantly because of how we play. We know he's comfortable playing short, he's just in a team where he's used as a playmaker at times for hitting space high up the pitch.

 

[sifi36]

I always thought it was completely insane that we just watched Rulli go to Ajax in January for £8m without making any kind of a move, the data confirms it.

It also confirms that we don't have to spend big to considerably improve on the goalkeeper, which I've been trying to tell people for months.

Thank you for putting it in a clear, readable, easy format. Great work.

You’re welcome.

I like Rulli from the data, though his ball comfort score suggests he is somewhat overconfident on the ball and prone to miscontrols and dispossessions. He’d get eaten alive if he gave the ball away cheaply early doors at United.

EDIT: On Raya, his short and medium accuracy are towards the bottom of the list (98% for each) and is independent of how may long balls he plays. Long accuracy is judged separately and for that he is mid-pack, though I can see how Brentford's style has them choosing to go long from more situations would negatively affect this. His "Passing Tendencies" score will certainly be affected by Brentford's style though.

 

[top1whoisman]

I always thought it was completely insane that we just watched Rulli go to Ajax in January for £8m without making any kind of a move, the data confirms it.

Not sure how Ajax fans feel about him but he made a lot of costly and non-costly mistakes in his first few appearances at Ajax. Looked like a horrible signing. Not sure if he’s improves since.


@BrilliantOrange
@Daslogisch
@Terranova
@AjaxCunian

 

[sifi36]

Not sure how Ajax fans feel about him but he made a lot of costly and non-costly mistakes in his first few appearances at Ajax. Looked like a horrible signing. Not sure if he’s improves since.

Ball playing errors or shot stopping ones? The former can be seen in the data, the latter could be explained by the fact that the data sample for Rulli is quite small so it’s possible he was on a hot streak shot stopping wise before the move.

 

[top1whoisman]

Ball playing errors or shot stopping ones? The former can be seen in the data, the latter could be explained by the fact that the data sample for Rulli is quite small so it’s possible he was on a hot streak shot stopping wise before the move.

Both if I remember correctly. Ajax really needed a keeper and I also thought they had done really well to get him for €8m but it was awkward watching him. Physically on another level to their veteran keepers and surely an upgrade but he’s been far from convincing.

 

[OleGunnar20]

Great stuff, very interesting. Costa looks great.

 

[JB7]

Not sure how Ajax fans feel about him but he made a lot of costly and non-costly mistakes in his first few appearances at Ajax. Looked like a horrible signing. Not sure if he’s improves since.

I haven't seen him at Ajax to be fair. Look I'm not saying he's a top top goalkeeper, but he is an improvement on what we have and was available for a very reasonable fee. In a world where we need to spend big on a striker, bringing in a lower cost goalkeeper was always going to be a good idea and for £8m Rulli should've been a no brainer, somewhat like Sommer will be if he's available for £5m again in the summer, or Raya if he's available for £25m or so.

Samba is very interesting, I know a few Forest fans and they absolutely loved him before his move last summer and he's clearly performed very well this season too at a higher level.

We watched Maignan move clubs for £15m & Donnarumma for nothing in the same summer which I'm still not over in truth.

 

[sokol11]

I've been analysing the goalkeeper data available through FBRef as I think goalkeeping is the easiest of all positions to analyse statistically.

Fantastic read. Didn't open the file to see if they are there. But out of curiosity where does Ramsdale rank or Pope? They had good seasons but not in the top 50 in the list?

 

[top1whoisman]

I haven't seen him at Ajax to be fair. Look I'm not saying he's a top top goalkeeper, but he is an improvement on what we have and was available for a very reasonable fee. In a world where we need to spend big on a striker, bringing in a lower cost goalkeeper was always going to be a good idea and for £8m Rulli should've been a no brainer, somewhat like Sommer will be if he's available for £5m again in the summer, or Raya if he's available for £25m or so.

Samba is very interesting, I know a few Forest fans and they absolutely loved him before his move last summer and he's clearly performed very well this season too at a higher level.

We watched Maignan move clubs for £15m & Donnarumma for nothing in the same summer which I'm still not over in truth.

Based on what I’ve seen Rulli is not a better goalkeeper than DDG. If he was I’m sure plenty of clubs would’ve been after him in January and the fee would’ve been higher as a result.

 

[captaincantona]

it’s Keylor Navas all day isn’t it. Is still a top keeper. Cost next to nothing and allow us to focus on other areas for the next few windows.

 

[Trumpeter Whydah]

Quality post @sifi36 and it highlights what optics tell for a while now this season, Costa looks the next big one. Personally, I'd factor in a coefficent > 1 for Maignan, since SerieA this year was really challenging for keepers, again optically - so I'll ask you this: how do stats look for Costa when factoring only Europe and top half of the table?

 

[JB7]

Based on what I’ve seen Rulli is not a better goalkeeper than DDG. If he was I’m sure plenty of clubs would’ve been after him in January and the fee would’ve been higher as a result.

Fair enough, that's your view, I've seen him several times and would swap him for De Gea in a heartbeat. Unless he's lost an arm since joining Ajax, anyway.

You'd have thought clubs would have been after several of the goalkeepers who have moved clubs for reasonable fees in the past few years, it's an odd one.

 

[sifi36]

Fantastic read. Didn't open the file to see if they are there. But out of curiosity where does Ramsdale rank or Pope? They had good seasons but not in the top 50 in the list?

Both are somewhat impacted by playing for Burnley and Sheffield United in previous seasons.

Pope can be seen in the second image, focussed on stopping and cross claiming. His scores for both passing metrics are bad, dragging him out of the top 50 on the more balanced table.

Arsenal don’t really use Ramsdale in build up all that much and he’s made quite a few stopping and ball-playing errors this season and more so last season (see Southampton). He’s kind of average overall.

 

[Born2Lose]

Based on what I’ve seen Rulli is not a better goalkeeper than DDG. If he was I’m sure plenty of clubs would’ve been after him in January and the fee would’ve been higher as a result.

How many clubs are after De Gea now he's technically available on a free?

 

[top1whoisman]

How many clubs are after De Gea now he's technically available on a free?

If his wage demands weren’t that high I’m sure there’d be quite a few clubs in Europe interested in a solid, experienced keeper.

 

[Eric_the_Red99]

I knew DDG was bad at crosses but holy feck. Just 5%?!

Absolutely agree that upgrading can be done inexpensively. And I think it should be done inexpensively tbh, as we can get far more bang for our buck elsewhere on the pitch.

But we definitely do need to upgrade. I’d say no regular first team starter is currently holding us back to the extent that DDG is, so getting a replacement in for next season should be a no brainer.

 

[Oranges038]

I've been analysing the goalkeeper data available through FBRef as I think goalkeeping is the easiest of all positions to analyse statistically.

I've taken some of the provided ones and come up with a few transformations to create a list of metrics to score goalkeepers from the top five leagues on their league performances over this season and the previous two to create reasonable size samples.

Metric	Definition	Limitations
Stopped Crosses %	Crosses Stopped divided by Crosses Faced	Doesn't take into account how "stoppable" a cross is
Distance Swept /90	Number of sweeping actions multiplied by average sweeping distance, divided by 90s played
Shot Stopping	Post-shot xG faced minus goals against, divided by post-shot xG faced	FBRef post-shot xG model doesn't take into account shot location or power
Bad Touches /90	Miscontrols plus dispossessions per 90 played
Bad Touch Ratio	Miscontrols plus dispossessions divided by total touches
Outside Area Touches	Touches Outside penalty area divided by total touches
Outside Area Touches /90	Touches Outside penalty area per 90 played
Short and Medium Passes /90		Somewhat team style dependent
Short and Medium Share	A ratio of short and medium passes relative to total passes	Somewhat team style dependent
Short Accuracy	Completion percentage of short passes
Medium Accuracy	Completion percentage of medium passes
Long Accuracy	Completion percentage of long passes
Passing Tendancies	An average of the percentile rank of "Short and Medium Passes /90" and "Short and Medium Share"
Passing Accuracy	An average of the percentile rank of "Short Accuracy", "Medium Accuracy" and "Long Accuracy"
On Ball Comfort	An average of the percentile rank of "Bad Touches /90", "Bad Touch Ratio", "Outside Area Touches" and "Outside Area Touches /90"
Crosses	Percentile rank of "Stopped Crosses %"
Sweeping	Percentile rank of "Distance Swept /90"
Shot Stopping	Percentile rank of "Shot Stopping"
OVERALL SCORE UNWEIGHTED	An Average of each of the six above
OVERALL SCORE WEIGHTED	As above but each category is weighted by the amount specified

That then gives us the below table for the top 50 overall (Costa is highlighted as he doesn't play in a top five league and thus his data may or may not be comparable):

Lots of names you would expect, but also quite a few that I hadn't heard of: Casteels (contract expires 2024 and valued at 8m Euros on TransferMarkt), Samba (10m Euros) and Diouf (5m Euros) look very well balanced, with none of the six metrics being below average. Diogo Costa, Pau Lopez and to a lesser extent David Raya also look like good options for United.

David De Gea for comparison:

If you're more of a traditionalist and value shot-stopping and command of area uber-alles, quadrupling the weight of those two metrics gives the following top 50

For a visual comparison of some we've been linked to and others we should be considering:

Conclusions

• David De Gea is either average or poor at everything a goalkeeper is required to do
• Upgrading on him can be done in an inexpensive manner
• Obviously scouting/eye-test and personality assessment would also be required before signing any of the above

You can download the Excel file from the link below and add your own weightings for each category in the first sheet ("Scoreboard"), I'd recommend filtering "90s" to at least 30 if not 50 to ensure the goalkeepers being looked at have reasonable sample sizes. The second ("Comparison Charts") allows you to create your own radar comparison chart for up to five of the goalkeepers included in the analysis.

Link to File

Excellent work, shows any number of keepers that could be brought in for a reasonable fee and it would still improve the team.

WOW! DDG has been that bad over the last 3 seasons!

From the table Costa looks like a no brainer.

He has.

Some of us tried to say it, but "saves" you know.

 

[sifi36]

it’s Keylor Navas all day isn’t it. Is still a top keeper. Cost next to nothing and allow us to focus on other areas for the next few windows.

He’s an excellent stopper, but is as bad as Dave on crosses and not much better as a sweeper or with his feet.

There’s better options out there in my view.

 

[ifightdragons]

This is a fantastic round-up, showing everyone exactly how poor of a goalkeeper David de Gea truly is...

It's borderline criminal that he is the joint highest paid footballer in the Premier League with Haaland, and the highest paid goalkeeper in the world, ever.

What worries me is why United haven't replaced him before.
It genuinely scares me that they thought giving him a new contract on those wages was anywhere near justified.
And it downright frightens me that they now are wanting him to stay, even on reduced wages.
Surely, they can't be looking at any relevant data, if this is and has been their idea of making good decisions?

I wouldn't want a keeper like that for free at any club that is supposed to challenge the likes of Pep.

On a side note:
People like to say that he used to be great from 2016-2018.
He was still just as incredibly poor at everything except shot stopping back then.
In the 2013-14 season I realized that he is scared of physical duels, can't pass through a moderate press, won't sweep or command his area.

Admittedly, he was a world class shot stopper... But anyone who has kept up with the evolution of football, realized by then that a shot stopper simply isn't enough

The cold, hard truth:
David de Gea was always a bad goalkeeper and footballer.
He was just a great shot stopper.

 

[onemanarmy]

Fair enough, that's your view, I've seen him several times and would swap him for De Gea in a heartbeat. Unless he's lost an arm since joining Ajax, anyway.

You'd have thought clubs would have been after several of the goalkeepers who have moved clubs for reasonable fees in the past few years, it's an odd one.

Rulli has been a disaster at Ajax.

 

[sifi36]

Quality post @sifi36 and it highlights what optics tell for a while now this season, Costa looks the next big one. Personally, I'd factor in a coefficent > 1 for Maignan, since SerieA this year was really challenging for keepers, again optically - so I'll ask you this: how do stats look for Costa when factoring only Europe and top half of the table?

Unfortunately, the data is only available for the season as a whole, rather than game-by-game. League difficulty is a tricky one, I'm not sure there's an easy way to tease that out without more of the underlying event-level data (which costs tens of thousands of Pounds). Comparing Costa's record in the Champions League against other keepers in their domestic leagues would be unfair to him; the other keepers would have Southampton or Sampdoria at home padding their records out.

Rulli has been a disaster at Ajax.

Looks like his performances have generally been OK, bar some high-profile, mistake-strewn stinkers against Union Berlin, PSV and Feyenoord. Would that be correct to say?

The data suggests that he makes on-ball errors (miscontrols and disposssessions) very regularly, around the same rate as Meslier and Kepa - so in the bottom 20 out of 198 goalkeepers in the analysis.

 

[MadDogg]

Good thread, and shows what many of us have been saying for years.

WOW! DDG has been that bad over the last 3 seasons!

It's been more than 3 seasons unfortunately. More like 5.

 

[ifightdragons]

Good thread, and shows what many of us have been saying for years.


It's been more than 3 seasons unfortunately. More like 5.

More like 10 seasons, or even yet, his entire career.

The only thing that's really changed is his shot stopping.

Everything else has or more less remained constant, which is to say, relatively poor.

Except for his declining shot stopping, he is more or less the exaxct same keeper now as he was when he first joined.

Most football fans didn't start seeing this, before Allison and Ederson joined the PL (around 5 years ago, give or take).
Before them, most fans didn't evaluate De Gea in comparison to modern goalkeepers.
But truth be told, keepers like Neuer and Van der Sar were 10-15 years ahead of the curve, and De Gea was already miles behind when his career started.

Even in his "best" seasons, he was still a huge detriment to us, and especially under van Gaal, who wanted to implement the principles of Total Football with a squad not fit for purpose. He was then a rather suitable goalkeeper for counter-attacking Mourinho, and finally declined rapidly under Ole after the 2018 World Cup -- despite being a decent match for Ole's counter-attacking style.
His decline has been made ever-so apparent again, seeing as how Ten Hag now wants to implement his brand of Total Football, but simply had to give up playing out from the back, or maintaining a high line, only 2 games into the season.

So he has really been a poor goalkeeper his entire career. It was just masked for years by people being blinded by his saves, and not understanding what a modern goalkeeper is.

 

[CantonaManc]

Somebody needs to send this to ETH and the board

Great stuff anyway, I hope they don't renew De Gea and we go for Costa in the summer or anybody at this point.

 

[pogbasformerbarber]

The only weird anecdotal name on this list is Kepa, who strikes me as poor when I watch him play. Maybe I'm just catching him on his worst days...

Cotsa is a crucial buy this summer IMHO. He's amazing to watch live, and this table seems to support that. Theres enough smoke that we are heavily scouting him, so I think the club is sold on him if we can get the funds sorted with the take-over.

I've watched a fair amount of Fulham games this season and the difference in quality of Leno vs. De Gea is instantly palpable. On a side, what an amazing bit of business they did this summer with Leno, Palhinha, and Willian. But to the point, Fulham seems to have some ambition so I doubt Leno would directly be an option, but it's been clear to me that we desperately need an upgrade.

 

[Zlatattack]

I've been analysing the goalkeeper data available through FBRef as I think goalkeeping is the easiest of all positions to analyse statistically.

I've taken some of the provided ones and come up with a few transformations to create a list of metrics to score goalkeepers from the top five leagues on their league performances over this season and the previous two to create reasonable size samples.

Metric	Definition	Limitations
Stopped Crosses %	Crosses Stopped divided by Crosses Faced	Doesn't take into account how "stoppable" a cross is
Distance Swept /90	Number of sweeping actions multiplied by average sweeping distance, divided by 90s played
Shot Stopping	Post-shot xG faced minus goals against, divided by post-shot xG faced	FBRef post-shot xG model doesn't take into account shot location or power
Bad Touches /90	Miscontrols plus dispossessions per 90 played
Bad Touch Ratio	Miscontrols plus dispossessions divided by total touches
Outside Area Touches	Touches Outside penalty area divided by total touches
Outside Area Touches /90	Touches Outside penalty area per 90 played
Short and Medium Passes /90		Somewhat team style dependent
Short and Medium Share	A ratio of short and medium passes relative to total passes	Somewhat team style dependent
Short Accuracy	Completion percentage of short passes
Medium Accuracy	Completion percentage of medium passes
Long Accuracy	Completion percentage of long passes
Passing Tendancies	An average of the percentile rank of "Short and Medium Passes /90" and "Short and Medium Share"
Passing Accuracy	An average of the percentile rank of "Short Accuracy", "Medium Accuracy" and "Long Accuracy"
On Ball Comfort	An average of the percentile rank of "Bad Touches /90", "Bad Touch Ratio", "Outside Area Touches" and "Outside Area Touches /90"
Crosses	Percentile rank of "Stopped Crosses %"
Sweeping	Percentile rank of "Distance Swept /90"
Shot Stopping	Percentile rank of "Shot Stopping"
OVERALL SCORE UNWEIGHTED	An Average of each of the six above
OVERALL SCORE WEIGHTED	As above but each category is weighted by the amount specified

That then gives us the below table for the top 50 overall (Costa is highlighted as he doesn't play in a top five league and thus his data may or may not be comparable):

Lots of names you would expect, but also quite a few that I hadn't heard of: Casteels (contract expires 2024 and valued at 8m Euros on TransferMarkt), Samba (10m Euros) and Diouf (5m Euros) look very well balanced, with none of the six metrics being below average. Diogo Costa, Pau Lopez and to a lesser extent David Raya also look like good options for United.

David De Gea - 78th out of 90 goalkeepers that have played at least 50 games' worth of minutes:

If you're more of a traditionalist and value shot-stopping and command of area uber-alles, quadrupling the weight of those two metrics gives the following top 50

For a visual comparison of some we've been linked to and others we should be considering:

Conclusions

• David De Gea is either average or poor at everything a goalkeeper is required to do
• Upgrading on him can be done in an inexpensive manner
• Obviously scouting/eye-test and personality assessment would also be required before signing any of the above

You can download the Excel file from the link below and add your own weightings for each category in the first sheet ("Scoreboard"), I'd recommend filtering "90s" to at least 30 if not 50 to ensure the goalkeepers being looked at have reasonable sample sizes. The second ("Comparison Charts") allows you to create your own radar comparison chart for up to five of the goalkeepers included in the analysis.

Link to File

Posts like this deserve some sort of like button. Proper top quality.

 

[Annihilate Now!]

Costa does seem to be one - but I can't shake how rubbish he looked for Portugal at the World Cup.

Never even heard of Yehvann Diouf before... stats and age seem a bit too good to be true. He is only 6ft 1 though which is a tiny bit on the short side for a keeper.

Always feel like a keeper should be, minimum 6ft3.... (why I'm a bit hesitant on Raya) but then I am a height supremacist.

 

[Teja]

Great thread op! I hope people stay away from the DDG bashing in the thread and focus on the data and how OP arrived at their conclusions.

Some questions

On Ball Comfort

An average of the percentile rank of "Bad Touches /90", "Bad Touch Ratio", "Outside Area Touches" and "Outside Area Touches /90"

I assume you have to use per 90 numbers for all of these and not the raw numbers. Why use the raw values at all?

I'm promptly going to contradict myself and say it seems like the raw average distance of defensive action is a better metric than outside area touches etc.?

Are things like carries, progressive passing distance not valuable?

Diogo Costa

How much of Costa's stats are being inflated by playing in a good team?

 

[Teja]

Never even heard of Yehvann Diouf before... stats and age seem a bit too good to be true. He is only 6ft 1 though which is a tiny bit on the short side for a keeper.

This also keeps the statisticians modest. Otherwise they get drunk by their own power to scout. There will always be randos who happen to score well in your particular model that aren't very good.

(not saying Yehvann Diouf isn't, I have no idea who he is)

But not having guys like Ederson, Ramsdale etc. higher up doesn't pass the sniff test. Maybe smaller samples need to get downweighted more and you need a per-league coefficient to adjust for quality. Lots of ways to cook the sausage. Raw numbers only go so far.

 

[croadyman]

Costa does seem to be one - but I can't shake how rubbish he looked for Portugal at the World Cup.

Never even heard of Yehvann Diouf before... stats and age seem a bit too good to be true. He is only 6ft 1 though which is a tiny bit on the short side for a keeper.

Always feel like a keeper should be, minimum 6ft3.... (why I'm a bit hesitant on Raya) but then I am a height supremacist.

I remember that one error in World Cup against Ghana but nothing else stood out

 

[Mmm-Qatarian]

Costa does seem to be one - but I can't shake how rubbish he looked for Portugal at the World Cup.

Never even heard of Yehvann Diouf before... stats and age seem a bit too good to be true. He is only 6ft 1 though which is a tiny bit on the short side for a keeper.

Always feel like a keeper should be, minimum 6ft3.... (why I'm a bit hesitant on Raya) but then I am a height supremacist.

Unacceptable heightism from a scout. Ban please.

In all seriousness, I also generally prefer a big bastard in goal but on my end I think it's just entirely irrational. I wouldn't really be opposed to signing a (relatively) shorter one if they ticked all the boxes.

 

[SirAF]

Top post and analysis @sifi36!

 

[sifi36]

The only weird anecdotal name on this list is Kepa, who strikes me as poor when I watch him play. Maybe I'm just catching him on his worst days...

Cotsa is a crucial buy this summer IMHO. He's amazing to watch live, and this table seems to support that. Theres enough smoke that we are heavily scouting him, so I think the club is sold on him if we can get the funds sorted with the take-over.

I've watched a fair amount of Fulham games this season and the difference in quality of Leno vs. De Gea is instantly palpable. On a side, what an amazing bit of business they did this summer with Leno, Palhinha, and Willian. But to the point, Fulham seems to have some ambition so I doubt Leno would directly be an option, but it's been clear to me that we desperately need an upgrade.

Kepa scores very poorly on ball comfort (he makes a fair amount of errors in possession) and crosses (he's tiny for a keeper). Giving equal weight to each category prevents him from looking too bad, I think applying weights to some categories would not favour him.

Costa does seem to be one - but I can't shake how rubbish he looked for Portugal at the World Cup.

Never even heard of Yehvann Diouf before... stats and age seem a bit too good to be true. He is only 6ft 1 though which is a tiny bit on the short side for a keeper.

Always feel like a keeper should be, minimum 6ft3.... (why I'm a bit hesitant on Raya) but then I am a height supremacist.

This also keeps the statisticians modest. Otherwise they get drunk by their own power to scout. There will always be randos who happen to score well in your particular model that aren't very good.

(not saying Yehvann Diouf isn't, I have no idea who he is)

But not having guys like Ederson, Ramsdale etc. higher up doesn't pass the sniff test. Maybe smaller samples need to get downweighted more and you need a per-league coefficient to adjust for quality. Lots of ways to cook the sausage. Raw numbers only go so far.

Diouff has fewer than my recommended 50 games played (he only began starting this season). I'd describe him as one to watch over the next couple of seasons rather than a must-buy right now. In any case it looks like he's prone to errors in possession (see on ball comfort).

I would suggest that 6'1" is the perfect height

Ederson is 9th on the unweighted list and 5th if you filter out players with fewer than 50 90s, see post #5. Ramsdale is impacted by a season of playing for Sheffield United, plus he makes loads of mistakes and isn't really that good. Aside from sweeping, going long a lot and have penalty takers miss the goal, he doesn't rate particularly well on much, below are his stats for this season.

Great thread op! I hope people stay away from the DDG bashing in the thread and focus on the data and how OP arrived at their conclusions.

Some questions



I assume you have to use per 90 numbers for all of these and not the raw numbers. Why use the raw values at all?

I'm promptly going to contradict myself and say it seems like the raw average distance of defensive action is a better metric than outside area touches etc.?

Are things like carries, progressive passing distance not valuable?



How much of Costa's stats are being inflated by playing in a good team?

The images above are showing percentile ranks for each metric.

The underlying data is normalised per 90, unless it wouldn't make sense to do so, and can be seen in the linked Excel file.

For most of these, I've used per 90, unless it doesn't make sense to do so. No raw numbers have been used.

Crosses stopped per 90 would be impacted by the defence and midfield in front of the goalkeeper, whereas % of crosses stopped normalises for the number faced. For example, Ederson faces far fewer crosses than Bazunu. Crosses stopped per 90 would heavily favour Bazunu, even though they stop a similar ratio.

Equally the shot-stopping number would favour goalkeepers who face a lot of shots if it was per 90, so instead it's normalised for post-shot xG (a measure of quantity and difficulty of shots faced).

For some(Bad Touches and Short and Medium Passes), I've got per 90 and a ratio of total touches/passes. Having this ratio as well as the per 90 equivalents helps goalkeepers who don't see a lot of the ball due to team style or lack of team possession.

Outside Area Touches is more a measure of starting position and involvement in possession and feeds into on ball comfort rather than sweeping.

I looked at progressive passing distance, though because it only counts completed passes and disregards failed ones, it didn’t really show any real patterns that looked sensible. It made keepers who go long almost always look better than ones who go short and in a world of playmaking goalkeepers that feels like the wrong direction.

 

[Pavl3n]

Great post and analysis. Thanks mate.