xG - hype or valuable stat?

[ap04]

16 minutes ago, AndyinLiverpool said:

I don't understand this. Expected goals is more representative than actual goals? What are they more represetative of?

Of how many goals you deserve with the same quality, or how many you should score after 1000 games.

Same as if you roll a dice, the actual outcome will be anything, the expected value (3.5) is what you deserve to get.

[roboto]

5 hours ago, MadAmster said:

As an old school type, I obviously go for hype. Is it a more valuable stat than shots, shots on target? I don't think it is.

Here's some examples from last weekend

Luton 1.75 Sheff U 1.56. The result was 1-3
Villa 2.45 Utd 1.9. The result was 1-2
Red Dogs 1.25 Toon 0.68. The result was 2-3
Fulham 1.75 B'mouth 2.61. The result was 3-1

Shots/shots on were

Luton 20/5 Sheff U 7/3.
Villa 23/10 Utd 17/5.
Red Dogs 13/3 Toon 7/5.
Fulham 7/6 B'mouth 25/4.

Look at xG and, in all 4 cases the team with the better xG lost. Look at shots and the team with the most shots lost. Shots on target sees Luton and Villa losing despite more SOT and then Toon and Fulham having more SOT and winning.

IMO xG is no better a stat than shots/SOT. In the end, it all comes down to creating chances, making the decision to shoot and then whether it's off target, on target but saved/blocked or on target and in.

In the interest of balance, there were other games where the side with the best xG won.

I'm not a massive fan of xG and certainly wouldn't rely on it as the main indicator of performance. However, you've picked 4 of the 10 EPL fixtures from the weekend. Do the other 6 correlate with these 4? Let's look...

Crystal Palace 1.00 Chelsea 1.33. The result was 1-3 (xG predicted winner but was a little off with amount Chelsea scored.
West Ham 1.58 Arsenal 1.92. The result was 0-6 (again xG got the winner correct but was off with the scale of the score)
Liverpool 2.67 Burnley 0.67. The result was 3-1 (quite close this one)
Spurs 2.25 Brighton 1.00. The result was 2-1 (spot on)
Wolves 1.50 Brentford 0.91. The result was 0-2 (this one was off)
Man City 2.45 Everton 1.42. The result was 2-0 (correct winner, but a bit out with Everton)

I'm not sure which site you used for your xG results but the one I looked at also Fulham 1.83 Bournemouth 1.17, which was closer to the 3-1 result than your xG stats.

All in all, I'm not a fan of xG and we can see that there's discrepancy between different sites listing their stats. However, it's not fair to cherry pick half of the weekends results to satisfy your argument and not include the rest of the results to at least add context or back up your findings.

[AndyinLiverpool]

1 hour ago, ap04 said:

Of how many goals you deserve with the same quality, or how many you should score after 1000 games.

Same as if you roll a dice, the actual outcome will be anything, the expected value (3.5) is what you deserve to get.

To what end?

[McArthur Park]

1 hour ago, ap04 said:

Of how many goals you deserve with the same quality, or how many you should score after 1000 games.

Same as if you roll a dice, the actual outcome will be anything, the expected value (3.5) is what you deserve to get.

Probably being a bit thick, but why does it matter how many goals you deserve, or should score? What's the point? 

[Day]

5 hours ago, Ghost of Clough said:

But, what if you can't watch all 552 League 1 games to make a fair judgement? How do you compare how you're doing with other teams?

The league table is a good indicator of how teams are doing.

Football isn’t overly complex, stick the ball in the net more than the other team and you get 3 points. 

Not sure how else you can actually compare, seen other comments in this topic about what teams “deserve” to justify the xG argument, it’s all the same, ignoring the basic rules of the game which is based on actual goals.

[Tamworthram]

52 minutes ago, McArthur Park said:

Probably being a bit thick, but why does it matter how many goals you deserve, or should score? What's the point? 

I’m being even thicker. How can you “deserve” to get any particular number when rolling a dice, least of all a number (3.5j that it’s impossible to get? 🤷🏻‍♂️

[Ghost of Clough]

4 hours ago, David said:

The league table is a good indicator of how teams are doing.

Football isn’t overly complex, stick the ball in the net more than the other team and you get 3 points. 

Not sure how else you can actually compare, seen other comments in this topic about what teams “deserve” to justify the xG argument, it’s all the same, ignoring the basic rules of the game which is based on actual goals.

At what point is the league a fair indicator? Do we have to wait until the last game has been played?

10 games in, Port Vale were 6th bit now they're 4th bottom. After 20 games, Reading were bottom 4, but would now be 14th without a points deduction. Leyton Orient have shot up from 16th to 9th in the same period of time.

[Day]

2 minutes ago, Ghost of Clough said:

At what point is the league a fair indicator? Do we have to wait until the last game has been played?

10 games in, Port Vale were 6th bit now they're 4th bottom. After 20 games, Reading were bottom 4, but would now be 14th without a points deduction. Leyton Orient have shot up from 16th to 9th in the same period of time.

Every week. Clubs dip in and out of form, move up and down the table. Ultimately come May the teams over the season are ranked 1 to 24 where they deserve to be.

[DB83]

2 minutes ago, Ghost of Clough said:

At what point is the league a fair indicator? Do we have to wait until the last game has been played?

10 games in, Port Vale were 6th bit now they're 4th bottom. After 20 games, Reading were bottom 4, but would now be 14th without a points deduction. Leyton Orient have shot up from 16th to 9th in the same period of time.

Yes. The only way to know whether a team has had a successful season is to look at the table at the end of the season. For a team like, say, Cheltenham, if they stay up by even a goal scored, it will have been a successful season. If they go down by a goal scored it will have been a failure. For Derby, if we finish second it will have been a successful season. If we miss out, however narrowly, and then lose in the play-offs, most will consider the season a failure. The only thing that matters is the outcome at the very end of the season. Possession, shots on target or expected goals are irrelevant compared to final league position.

[Comrade 86]

0.98 - 1.27 FWIW

[bimmerman]

12 hours ago, Ghost of Clough said:

Give Chris Martin and Connor Sammon exactly the same chance and they'll have the same xG, but one of them will be more liekly to score as he's a better finisher. 

Huh? 

They wouldn't have the same xg cos Connor Sammon couldn't hit a cows arse with a banjo, so ones more likely to have a higher xg then Connor Sammon as surely his would be zero. Nonsense stat that doesn't make sense

Edited February 14 by bimmerman

[Crewton]

I understand the technical and statistical value of xG for coaches, analysts, gamblers etc, but I've never needed xG to tell me that x player couldn't hit a cow's a*se with a banjo.

 

[AndyinLiverpool]

Has anyone researched the success rate in gambling of expected goals over, say, tossing a coin or chucking a dart?

[MadAmster]

18 hours ago, roboto said:

I'm not a massive fan of xG and certainly wouldn't rely on it as the main indicator of performance. However, you've picked 4 of the 10 EPL fixtures from the weekend. Do the other 6 correlate with these 4? Let's look...

Crystal Palace 1.00 Chelsea 1.33. The result was 1-3 (xG predicted winner but was a little off with amount Chelsea scored.
West Ham 1.58 Arsenal 1.92. The result was 0-6 (again xG got the winner correct but was off with the scale of the score)
Liverpool 2.67 Burnley 0.67. The result was 3-1 (quite close this one)
Spurs 2.25 Brighton 1.00. The result was 2-1 (spot on)
Wolves 1.50 Brentford 0.91. The result was 0-2 (this one was off)
Man City 2.45 Everton 1.42. The result was 2-0 (correct winner, but a bit out with Everton)

I'm not sure which site you used for your xG results but the one I looked at also Fulham 1.83 Bournemouth 1.17, which was closer to the 3-1 result than your xG stats.

All in all, I'm not a fan of xG and we can see that there's discrepancy between different sites listing their stats. However, it's not fair to cherry pick half of the weekends results to satisfy your argument and not include the rest of the results to at least add context or back up your findings.

The last sentence of my post to which you replied reads... In the interest of balance, there were other games where the side with the best xG won. 

The problem  with xG is that the various sources use different statistical models and their own algorithms, hence the variance in the figures. Same goes for shots/SOT. The BBC stats are often different to those of DCFC or Sporting Life or "insert newspaper name of your choice".

I've not looked up last night's xG stats but I think our 3 goals wouldn't score well under xG but Collo's early, saved strike would. Conversely, they created a few chances inside our box but either mis-controlled or shot wide/over and those missed chances would probably attract a better xG than our actual goals.

The old adage of there are lies, damned lies and then there are statistics is as true as it ever was. Same with opinion polls, questions get formulated in order to virtually guarantee producing the required outcome.

Whatever any stats say about last night's game, the fact of the matter is that we set out with a game plan, executed it very well and won 3-0. I expect Stevenage to play similarly and to be very physical at Pride Park on Saturday and I hope we don't succumb to this type of game which is one we, historically, haven't been consistent in combatting.

[nottingram]

49 minutes ago, MadAmster said:

The last sentence of my post to which you replied reads... In the interest of balance, there were other games where the side with the best xG won. 

The problem  with xG is that the various sources use different statistical models and their own algorithms, hence the variance in the figures. Same goes for shots/SOT. The BBC stats are often different to those of DCFC or Sporting Life or "insert newspaper name of your choice".

I've not looked up last night's xG stats but I think our 3 goals wouldn't score well under xG but Collo's early, saved strike would. Conversely, they created a few chances inside our box but either mis-controlled or shot wide/over and those missed chances would probably attract a better xG than our actual goals.

The old adage of there are lies, damned lies and then there are statistics is as true as it ever was. Same with opinion polls, questions get formulated in order to virtually guarantee producing the required outcome.

Whatever any stats say about last night's game, the fact of the matter is that we set out with a game plan, executed it very well and won 3-0. I expect Stevenage to play similarly and to be very physical at Pride Park on Saturday and I hope we don't succumb to this type of game which is one we, historically, haven't been consistent in combatting.

There will always be variations though - the problem is as soon as you add more variables (defensive positioning, height of ball etc) your sample size from which the xG is drawn gets ever smaller, so there is a balance. Internal data within football clubs is likely to be pretty sophisticated though. 

Your last paragraph shows the pitfalls of using it for individual games. I see sometimes people say it is subjective but if anything the problem is that it is totally objective. A team can score early (ish) and then let the other team rack up low quality chances and they end up “losing” on xG. The reality is they might have been totally in control of the game. The longer term look at that might be that letting that happen every week will probably catch up on you when some of those low quality chances you’re happy to give up fly in to the top corner. 

 

[duncanjwitham]

2 hours ago, MadAmster said:

I've not looked up last night's xG stats but I think our 3 goals wouldn't score well under xG but Collo's early, saved strike would. Conversely, they created a few chances inside our box but either mis-controlled or shot wide/over and those missed chances would probably attract a better xG than our actual goals.

The first site I've looked at had xG at 1.24 v 1.24 for last night, and that basically lines up with your analysis there.  Both sides had a few decent chances.  We scored one of ours (the NML goal was probably a fairly good chance, xG-wise), missed our best chance (Collins), and scored a couple more from less-good chances.  They scored none of their chances (you literally described them as being "wasteful" in the match thread, so you clearly think they were capable of scoring from those chances).

That tells me xG is doing what it's supposed to.  It's trying to tell you who created the best chances in the game, nothing more, nothing less than that.  If xG had come out at 0.0-3.0 or something similar, then it's an entirely useless stat - we already have a goals-scored stat for that.  

[Comrade 86]

7 hours ago, AndyinLiverpool said:

Has anyone researched the success rate in gambling of expected goals over, say, tossing a coin or chucking a dart?

The bookies themselves rely on such analysis quite a lot. They are big subscribers. On the other side, Matthew Benham, amongst other sports bettors, also relied heavily on stat analysis. He used xG in it's earliest guises to trade on football on the betting exchanges and made enough doing so to fund Matchbook, his own betting exchange. Matchbook's success in turn, allowed him to buy Brentford. 

On a general note, I'm not going to involve myself in this debate as sides are clearly de-marked already, but I would say that the way some are assuming statistical analysis is used in sport is fundamentally inaccurate. It is meant to supplement and complement other tools, such as video and live scouting and even form analysis, not replace them wholesale.

Edited February 14 by Comrade 86

[ap04]

17 hours ago, Comrade 86 said:

0.98 - 1.27

3 hours ago, duncanjwitham said:

1.24 v 1.24

Extremely accurate on this occasion which doesn't happen every day. 4 chances to 5, relatively even game, good away performance, joke of a scoreline.

[Day]

How is it a joke of a scoreline? It's not a game of chances, it's a game of sticking the ball in the net. We did it, they didn't, their loss.

[Gee SCREAMER !!]

23 minutes ago, ap04 said:

Extremely accurate on this occasion which doesn't happen every day. 4 chances to 5, relatively even game, good away performance, joke of a scoreline.

We'd probably have won that by more if we'd stepped it up rather than soak it up for 20 minutes so your probably correct. They had two chances.  One from not clearing properly that they hit a post with and one that a pissed up non league 50 yr old with blinkers on would have put away.  As we scored the chances we had  and tested them on other occasions we both scored more and defended better.  That's a deserved win.