Chances conceded - quick analysis

Uptop

Well-Known Member
Joined
10 Dec 2012
Messages
3,102
Location
Far North
Seen some discussions on here regarding our defence and the quality/ quantity of chances we give away compared to other teams. Looked into some xPG data to do some basic analysis of this.

Basically this data tries to predict the chance a certain shot will be a goal based on a whole load of different parameters which gives each shot a xG value, a larger value indicates a higher probability of being a goal.

Probably plenty on here know much more about it than I do.

Originally I had hoped to make my own fancy tables but struggling to find any current data for this season that is publicly available, so relying on others for this basic comparison.

So to begin with this is every goal we have conceded in the league this season.

City Goals Conceded:

IxLHhST.png


From a very quick glance we can see the majority of our goals conceded are from relatively low value shots, indicating goalkeeping errors are leading to low probability shots going in or players are scoring great goals against us.

Now lets compare that to say Tottenham, one of the best defences in the league.

Tottenham Goals Conceded:
0amTMZF.png


This shows the majority of their goals are conceded from high value chances with few low value goals being conceded. This would appear to indicate when they concede it is due to defensive errors / good opposition attacks leading to high quality chances which leaves a keeper with not much chance of making a save.

This quick glance between the two would seem to indicate we give up less high quality chances than one of the best defences in the league. This would also suggest our defence is doing a good job on the whole and a better shot stopper in goal would make a big difference.

A quick comparison of the goals conceded by Willy and Bravo.

Willy:
gFuuNEf.png

Bravo:
jSfBNzt.png


This highlights how Willy's apparent better shot stopping ability has lead to fewer low value shots being conceded (although he has played fewer matches) compared to Bravo where nearly all goals conceded were lower value and should in theory be savable to some degree.

This doesnt show if we are giving away good chances that we simply save however. Again I will compare to Tottenham.

City Saves:
iqSphjv.png

Tottenham Saves:
fVKhCiA.png

Tottenham have more high and medium value shot saves which shows our defence gives up less chances than theirs with Loris being the main factor between the goals conceded, particularly for shots just inside the 18yard box where Bravo has conceded most of his goals yet Loris appears to save.

So in summary: We give away not only less chances in total but less high value chances than one of the best defences in the league with most of our goals conceded coming from low chance shots from around 12 yards away from goal which have since stopped since Willy has been in goal. This suggests either they are brilliant goals or a better keeper would manage to save them. In theory if Willy had played all season the majority of our goals conceded would have been saved if current trends were applied.

This does not account for the fact Bravo played when the team was in the process of learning new tactics which I believe lead to more chances from around the 12yard range, chances which have since been reduced, nor the positives he provides in possession of the ball.

That was a quick run through of the stats on our goals conceded, hopefully its interesting to some people. If anyone interprets things differently just say, that is just how things appeared to me. There is so much more that could be shown like individual games or comparisons to other teams but that would be so much to post at one time.

https://public.tableau.com/profile/...ueShotDashboard2016-17/PLShotDashboard2016-17 this is where I got the data from so have a look yourself if interested.
 
So as one of the most fundamental concepts in statistics is mean regression we can expect Bravo to be the best keeper in the league going forwards.
 
Sssooooooooo........Bravo's shite at saving shots then?

Correct, on average he saves 1 and lets 1 in. At Barca his stats were much better, almost 3.5 saves per goal over his last 2 full seasons with them. Don't know if he's lost all confidence or if its something else.
 
Here is a table that compares actual goals for and against with xG/xGA (expected goals for and against based on the rough quality of chances for and against).

To help read the table and looking at City first, we have actually scored 60 goals and let in 35 with a goal difference of 25. Based on xG/xGA we should have scored 65.9 and let in 26.2 with a goal difference of 39.7. Overall our actual goal difference is -14.7, call it -15, WORSE than on our expected goal difference based on quality of chances for and against. This is the worst in the league.

This tells us something we all know already and Pep bangs on about all the time and that is we have had problems in both boxes.

However, where it is interesting is when you compare it against the other teams. Our xG total is 65.9 which is by far and away the best in the league and is indicative we are the best attacking side. Our xGA total is 26.2 which is only a tad worse than the 2 best defensive sides United and Chelsea.

x1d361.jpg
 
Pep will know this and it will be sorted in the summer.

It backs up your general opinion when watching.

Simply not good enough.
 
So as one of the most fundamental concepts in statistics is mean regression we can expect Bravo to be the best keeper in the league going forwards.
Only if he actually is the best keeper in the league (that's his mean standard), and he's been incredibly unlucky so far. If his form so far really is his mean standard, then we shouldn't expect anything to change.
 

Don't have an account? Register now and see fewer ads!

SIGN UP
Back
Top
  AdBlock Detected
Bluemoon relies on advertising to pay our hosting fees. Please support the site by disabling your ad blocking software to help keep the forum sustainable. Thanks.