Jump to content

Perhaps we're not as good as we think? Yes it's XG


Carl Sagan

Recommended Posts

Well, I don't know about goals, but I expected a draw v DYS, wins at Millwall, Rotherham, and Bolton, and draws v Blades and Baggies, so I make that a points difference (against reality) of +6, which puts us on 34 points and a runaway lead!

 

Here's my shouldacouldawoulda league table:

 

 1.  Derby County   P16    Pts34

2.  Meh!

Link to comment
Share on other sites

  • Replies 74
  • Created
  • Last Reply
57 minutes ago, Tamworthram said:

It still amazes me that people (must be an entire army) are employed to count and "assess" every shot a team has and then calculate these meaningless stats. 

These stats aren't supposed to be taken on their own though. You're not supposed to throw a party because you're top of the xG chart.  They're supposed to tell you something that you can then go and look at in matches to see what's actually happening.

The fact we're scoring a decent number of goals, from a lot of shots, but have a relatively low xG suggests we're creating a lot of low-quality chances.  And if you go and look at the shot totals and shot zones (whoscored.com has these), we have the 2nd most shots in total in the Championship, and by far the most from outside of the box.  We also have one of the lowest shot counts from inside the 6 yard box.  That points to the same conclusion, we have lots of shots, but quite a few of them are pot shots from outside the area.  So you go and watch some matches and see why this is happening - and I suspect it will be the games against the likes of Bolton and Rotherham that sway it, where we struggle to break down massed defences and end up shooting from distance.

That's maybe not the best example of using these kind of stats, as the conclusion is probably fairly obvious anyway, but there will be plenty of other cases where it can be very useful.  I remember a previous manager (can't remember which) talking about a midfielder's yards-covered-per-match having dropped in the previous few matches.  That's something you might not pick up from just watching the game (as you can't be watching all 11 players all of the time), so you go and look at the game and see why.  Is he struggling a bit with fitness?  Are we struggling to get the ball to him, so he can't make forward runs?  Is he being lazy and not tracking back enough?  Are we playing too many long balls and bypassing midfield so he's not getting involved enough?  And once you know why it's happening, you can try and sort it - rest him if he's tired, tell him off if he's being lazy, tweak the midfield shape a bit, or whatever.  But without that stat telling you something might be off, the issue might never have been spotted (or at least spotted as quickly).

Link to comment
Share on other sites

1 hour ago, McLovin said:

I don’t want to get into THAT debate again but all I can say is don’t be quick to dismiss it. Carl made a similar post last year when Rowett was in charge when we were around second in the league before the collapse.

Just enjoy the ride and see where it takes us because it’s a lot more enjoyable than last year.

What I will say though is that we are conceding too many goals which could cost us long term. There’s only so many times that we can come from behind to win.

Anotger 30 times will do us fine

Link to comment
Share on other sites

56 minutes ago, cannable said:

It’s a statistical way of measuring performance, hardly nonsense 

Firstly, I'd argue that statistics are often nonsense.

Secondly, how can it automatically be assumed to be a realistic measure of performance? Isn't it just counting the number of shots on target you have from various positions on the pitch (it seems to even take into account "whether it was a headed shot"! Is that a good thing or a bad one? and who defines whether it was "big chance"?) v the number of similar shots (headers!) you've conceded?

It seems to measure the number of good goal scoring shots (agreed, a lot is generally a measure of good performance) but doesn't consider why you may not have scored. Was it a "big chance" (normally a good thing to create one but, not always, some are gifted to you) but you fluffed it (surely a measure of bad performance) or did the goal keeper make a good safe (neither a measure of good or bad performance from the attacking teams perspective).

Finally, even the title of the article is nonsense. The team that deserves to be top is, wait for it, the team with the most points not the one with the best net XG. 

Link to comment
Share on other sites

22 minutes ago, duncanjwitham said:

These stats aren't supposed to be taken on their own though. You're not supposed to throw a party because you're top of the xG chart.  They're supposed to tell you something that you can then go and look at in matches to see what's actually happening.

The fact we're scoring a decent number of goals, from a lot of shots, but have a relatively low xG suggests we're creating a lot of low-quality chances.  And if you go and look at the shot totals and shot zones (whoscored.com has these), we have the 2nd most shots in total in the Championship, and by far the most from outside of the box.  We also have one of the lowest shot counts from inside the 6 yard box.  That points to the same conclusion, we have lots of shots, but quite a few of them are pot shots from outside the area.  So you go and watch some matches and see why this is happening - and I suspect it will be the games against the likes of Bolton and Rotherham that sway it, where we struggle to break down massed defences and end up shooting from distance.

That's maybe not the best example of using these kind of stats, as the conclusion is probably fairly obvious anyway, but there will be plenty of other cases where it can be very useful.  I remember a previous manager (can't remember which) talking about a midfielder's yards-covered-per-match having dropped in the previous few matches.  That's something you might not pick up from just watching the game (as you can't be watching all 11 players all of the time), so you go and look at the game and see why.  Is he struggling a bit with fitness?  Are we struggling to get the ball to him, so he can't make forward runs?  Is he being lazy and not tracking back enough?  Are we playing too many long balls and bypassing midfield so he's not getting involved enough?  And once you know why it's happening, you can try and sort it - rest him if he's tired, tell him off if he's being lazy, tweak the midfield shape a bit, or whatever.  But without that stat telling you something might be off, the issue might never have been spotted (or at least spotted as quickly).

Fair enough but, makes you wonder how managers used to cope with assessing their teams performance and that of individuals in the past (I'd love to know what Brian Clough would have to say about this). You'd like to think a decent management team can see for themselves whether their team are creating (and scoring) enough chances and whether an individual seems to be off the pace recently.

Also, as you say, they should only be used as indicators (not for journalists to be suggesting team A should be above team B simply because they have a better XG ratio). 

Link to comment
Share on other sites

I have never understood the expected goals garbage. 

When I was a young lad going out drinking I expected to cop off every night. So I can happily say my expected number of girls I copped off with is circa 1000. 

The actual number mind is much much much lower.

But thanks to expected stats garbage i'll go with the higher figure now. Result!

 

Link to comment
Share on other sites

2 minutes ago, Tamworthram said:

Fair enough but, makes you wonder how managers used to cope with assessing their teams performance and that of individuals in the past (I'd love to know what Brian Clough would have to say about this). You'd like to think a decent management team can see for themselves whether their team are creating (and scoring) enough chances and whether an individual seems to be off the pace recently.

Also, as you say, they should only be used as indicators (not for journalists to be suggesting team A should be above team B simply because they have a better XG ratio). 

It's pretty simple - we can just do it better now than they did.  A managers job is all about making decisions (who to pick, who to buy, how to set the team up etc), and if they are able to have more information available to base those decisions on, then it's borderline criminal if they don't make use of it.  Plus it's pretty much an arms race - every other manager is potentially using these stats to inform his decisions, so you have to.  Brian Clough wasn't up against a league filled with managers using data analysts and sports scientists, he was using the same level of info as everyone else.

There are 11 players on the pitch for 90+ minutes - that's potentially 16.5+hours of individual performances to watch - there's no way a manager can re-watch all of that for every game, when we're playing Saturday-Tuesday-Saturday, on top of all of his other duties. If he can sit down and re-watch the game a couple of times with a checklist of all of the things that the stats have highlighted might be of interest, it's a much better use of their time.

And for what it's worth, the xG table that was posted mid-season last season was proven correct - we were overachieving based on scoring and conceding at freakishly high and low rates.  When those rates drifted back to normal, we started dropping down the table, and only a change of system late on turned it around.

Link to comment
Share on other sites

16 minutes ago, Paul71 said:

I have never understood the expected goals garbage. 

When I was a young lad going out drinking I expected to cop off every night. So I can happily say my expected number of girls I copped off with is circa 1000.

Not quite - it's based on the key indicators, like  - did I actually talk to any girls, did any of them let me buy them a drink, did I get on really well with any of them etc

if you expected to cop off but stood in the corner all night and didn't even look at anyone then xCO is very low

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...