Stats Central

[SteveV]

Josh Saunders objectively sucks.

I took a look at keeper stats (through the 4/24 games). I tried to normalize performance by looking at things on a per minute basis (i.e. looked at goals allowed per minute of play for each goalkeeper). The other thing I did was used shots allowed per minute as a proxy for how good a team's defenders are. (I believe this is actually shots on goal, since obviously there are no keeper stats for shots that aren't on goal). The thinking is that a good defense allows fewer shots on goal.

What this let me do was try to separate goals allowed in terms of what's possibly due to a bad defense in general and what's possibly due to a bad keeper. Wanna know what I found? You do? OK - I found that NYCFC actually puts Saunders in a pretty good spot. NYCFC is tied for 3rd in shots allowed per minute of play - this is due either to NYC's stellar defense (?) or, maybe more likely, to the fact that they've actually possessed the ball pretty well.



And do you know how Saunders thanks his team? He ranks 17th in the league in goals allowed per minute of play (note: I combined 2 Dallas keepers in to one generic keeper since they each had significant minutes).



Now, I guess it's possible that NYCFC's defenders just happen to give up a higher % of "easy" shots (so that even though they give up fewer shots, the ones they give up are from spots where Saunders couldn't possibly save it). But it could also just be that Saunders sucks at saving shots on goal (he's 19th in the league with a 59% save percentage) - and isn't that just a more fun explanation?

OK, the final piece of the puzzle is to understand what the impact is on our results. To check that, I assumed the same NYCFC possession/defense (the same shots allowed per minute), and just adjusted Saunders save percentage from his horrible 59% up to the league average 72% (not even great - just assume he was an AVERAGE keeper).

If Josh Saunders was an average keeper, NYCFC would've given up 8 goals so far instead of 12 goals. 4 fewer goals in 7 matches, or 0.6 goals per match. NYCFC scores 1.3 goals per match and allows 1.7 goals per match (net -0.4 per match). If we had an average keeper instead of Saunders, this would flip to 1.3 goals for and 1.1 goals against per match (net +0.2 per match).

(P.S. - What do you all think of a "Stats Central" thread for statistical fun like this? Just didn't seem to fit in anywhere else, and I like the idea of being able to find all this type of stuff in one spot.)

 

[Gotham Gator]

Good stuff. It's been a rough year for Josh. I think he was actually very good at saving our bad defensive bacon last year, but this year, he is a step down... maybe two steps down.

And I like the idea of a thread featuring off-kilter stats, although I also like just doing a separate thread for each one.

 

[Seth]

Josh Saunders objectively sucks.

I took a look at keeper stats (through the 4/24 games). I tried to normalize performance by looking at things on a per minute basis (i.e. looked at goals allowed per minute of play for each goalkeeper). The other thing I did was used shots allowed per minute as a proxy for how good a team's defenders are. (I believe this is actually shots on goal, since obviously there are no keeper stats for shots that aren't on goal). The thinking is that a good defense allows fewer shots on goal.

What this let me do was try to separate goals allowed in terms of what's possibly due to a bad defense in general and what's possibly due to a bad keeper. Wanna know what I found? You do? OK - I found that NYCFC actually puts Saunders in a pretty good spot. NYCFC is tied for 3rd in shots allowed per minute of play - this is due either to NYC's stellar defense (?) or, maybe more likely, to the fact that they've actually possessed the ball pretty well.



And do you know how Saunders thanks his team? He ranks 17th in the league in goals allowed per minute of play (note: I combined 2 Dallas keepers in to one generic keeper since they each had significant minutes).



Now, I guess it's possible that NYCFC's defenders just happen to give up a higher % of "easy" shots (so that even though they give up fewer shots, the ones they give up are from spots where Saunders couldn't possibly save it). But it could also just be that Saunders sucks at saving shots on goal (he's 19th in the league with a 59% save percentage) - and isn't that just a more fun explanation?

OK, the final piece of the puzzle is to understand what the impact is on our results. To check that, I assumed the same NYCFC possession/defense (the same shots allowed per minute), and just adjusted Saunders save percentage from his horrible 59% up to the league average 72% (not even great - just assume he was an AVERAGE keeper).

If Josh Saunders was an average keeper, NYCFC would've given up 8 goals so far instead of 12 goals. 4 fewer goals in 7 matches, or 0.6 goals per match. NYCFC scores 1.3 goals per match and allows 1.7 goals per match (net -0.4 per match). If we had an average keeper instead of Saunders, this would flip to 1.3 goals for and 1.1 goals against per match (net +0.2 per match).

(P.S. - What do you all think of a "Stats Central" thread for statistical fun like this? Just didn't seem to fit in anywhere else, and I like the idea of being able to find all this type of stuff in one spot.)

I like this sort of analysis but there are two issues. One is that the goal from the previous match where Iraola was behind his man was not Saunders' fault. Could he maybe have stopped it? Perhaps, but most keepers wouldn't have been able to most of the time. So it was a shot on goal and not a save but that won't be reflected in your numbers.

The other issue is that there are shots the might've been stoppable and some that just simply aren't going to be stopped by anyone no matter who they are. In other words, these numbers aren't accounting for offensive skill.

I like the idea here though, and perhaps we can take these stats going forward and maybe adjust them to reflect made saves vs missed saves/poor play.

 

[SteveV]

I like this sort of analysis but there are two issues. One is that the goal from the previous match where Iraola was behind his man was not Saunders' fault. Could he maybe have stopped it? Perhaps, but most keepers wouldn't have been able to most of the time. So it was a shot on goal and not a save but that won't be reflected in your numbers.

The other issue is that there are shots the might've been stoppable and some that just simply aren't going to be stopped by anyone no matter who they are. In other words, these numbers aren't accounting for offensive skill.

I like the idea here though, and perhaps we can take these stats going forward and maybe adjust them to reflect made saves vs missed saves/poor play.

Yeah, I know, and that was addressed in the post:

Now, I guess it's possible that NYCFC's defenders just happen to give up a higher % of "easy" shots (so that even though they give up fewer shots, the ones they give up are from spots where Saunders couldn't possibly save it). But it could also just be that Saunders sucks at saving shots on goal (he's 19th in the league with a 59% save percentage) - and isn't that just a more fun explanation?​

Partly it's just fun to dump on Saunders. But I also don't really know of any statistical way to account for things that someone subjectively thinks was someone else's "poor play" (I think that was on Iraola too, but I saw others posting why they thought it wasn't his fault). And, to be fair, if you do that you have to go through every goal against every other goalkeeper and do the same! Honestly, it all probably just washes out with a large enough sample size.

 

[mgarbowski]

I'm with SteveV on the idea that bad luck and defensive lapses should wash out across the league. I think it's fair to assume that given enough games, each goalie will be subject to those in roughly similar proportions. I'm also sure there are outliers and exceptions, but until someone makes an effort to prove Saunders has especially poor luck, I'm content to go with the assumption he's in the heavy clustered part of the bell curve.

 

[Ulrich]

Can't forget that there have been at least 3 for absolute certain goals this season that would be charged to Saunders except his defense cleared them off the line. So he's even worse than the stats suggest.

 

[Midas Mulligan]

Can't forget that there have been at least 3 for absolute certain goals this season that would be charged to Saunders except his defense cleared them off the line. So he's even worse than the stats suggest.

Also, you can't discount how much danger he himself creates with poor ball skills and passing. This analysis may understate the improvement in goals allowed by a significant margin because there is no adjustment to our shots allowed that are due solely to Saunders technical and tactical deficiencies.

 

[Tom in Fairfield CT]

What stats account for the three times that his U-9 ball skills have directly resulted in a very good chance on goal, or the gentle touch he put on the ball to guide it to Larin's head?

 

[SteveV]

Here's a look at NYCFC strengths and weaknesses. I just pulled team stats from a couple sources (MLSSoccer.com and WhoScored.com) and converted everything to a per game basis. Then compared NYCFC to the average MLS team, where average is 1.0.

 

[mgarbowski]

Here's a look at NYCFC strengths and weaknesses. I just pulled team stats from a couple sources (MLSSoccer.com and WhoScored.com) and converted everything to a per game basis. Then compared NYCFC to the average MLS team, where average is 1.0.

View attachment 4800

Can you clarify how you handled different stats where sometimes you want a high number and in other you don't and whether that puts you above or below the line? Another way to ask? Does being above the line mean you are above league average in raw numbers or subjectively.

For example, in Shots Conceded you show us above league average. Does that mean we concede more shots than most or that we concede fewer which could be viewed as above average? And were you consistent all the way across?

 

[SteveV]

I adjusted things so it's set up to be consistent in that higher on chart = better than average for that stat. Just wanted to be able to glance at it and compare across the board where we're good and bad.

It's really just NYCFC's per game stat divided by the average per game stat for most stats (since for most, higher is better, like possession). But when there's a stat where a lower number is better (like shots allowed), I inverted before plotting; so 1/(NYCFC/average).

You should read it like "shots per game is about 32% better than average" and "goals against is about 18% worse than average, because Saunders' save % sucks vs. the average keeper".

 

[SteveV]

Fixed it to really just show % better or worse vs. average...

 

[mgarbowski]

So we are caught offside much less than average? It seems like Villa does it all the time. That's certainly his reputation. But thinking on it now I guess it has happened a lot less recently, and nobody else on our club seems to get caught with any consistency.

 

[Ulrich]

So we are caught offside much less than average? It seems like Villa does it all the time. That's certainly his reputation. But thinking on it now I guess it has happened a lot less recently, and nobody else on our club seems to get caught with any consistency.

Villa could get caught offsides in a bunch of those games because he was the one dribbling and he didn't trust the wingers enough to pass.

After the last two games, we may see the number of our offsides start to go up.

 

[FootyLovin]

Fixed it to really just show % better or worse vs. average...

View attachment 4803

What are FC/game and FS/game?

 

[SteveV]

What are FC/game and FS/game?

"Fouls committed" (by us) and "fouls suffered" (against us)

 

[Tom in Fairfield CT]

Think you know how possession in soccer is calculated? Someone with a stopwatch counts up the time each team has the ball, right?

https://chanceanalytics.com/2017/01/15/are-we-calculating-possession-all-wrong/

Think again. The number we see have nothing to do with time, and everything to do with counting possessions and dividing by the total number.

Throw out every assumption or conclusion you have ever made based on possession, since its nothing more than a passing stat. The guy in the article does it by time, and gets wildly different figures than the OPTA data showed.

Makes sense in our case since there have been times we've gotten dominated yet still had more possession. This pisses me off way more than it should.

 

[mgarbowski]

Think you know how possession in soccer is calculated? Someone with a stopwatch counts up the time each team has the ball, right?

https://chanceanalytics.com/2017/01/15/are-we-calculating-possession-all-wrong/

Think again. The number we see have nothing to do with time, and everything to do with counting possessions and dividing by the total number.

Throw out every assumption or conclusion you have ever made based on possession, since its nothing more than a passing stat. The guy in the article does it by time, and gets wildly different figures than the OPTA data showed.

Makes sense in our case since there have been times we've gotten dominated yet still had more possession. This pisses me off way more than it should.

So the first reason that soccer stats don't tell us as much as in other sports is that the stats don't measure what they claim to.

 

[Sabo]

Even if you do have accurate possession stats, they can be relatively meaningless. NYCFC year 1 possession was disproportionately knocking the ball around aimlessly around the back 4 and GK before somebody kicked it long to Villa. NYCFC year 2 possession was much more positive and attacking as PV forced us to play out of the back to generate offense. The same possession time among our defense was much more focused on generating an attack and chances -- there is no measurement (I know of) for that difference in quality of possession.

 

[Tom in Fairfield CT]

Even if you do have accurate possession stats, they can be relatively meaningless. NYCFC year 1 possession was disproportionately knocking the ball around aimlessly around the back 4 and GK before somebody kicked it long to Villa. NYCFC year 2 possession was much more positive and attacking as PV forced us to play out of the back to generate offense. The same possession time among our defense was much more focused on generating an attack and chances -- there is no measurement (I know of) for that difference in quality of possession.

Well if you are accurately reporting it, you could do time of possession in your own half, vs opponents half, etc. It all comes down to accurate data. This way of reporting possession and calling it "possession" is offensive to my brain.