xG table so far this season
- Thread starter AytonMews
- Start date
Howay Laughing, surely you see the that is blatant hypocrisy.
So it's not ok for me to challenge the xG over a match as being a false indicator but in the very same thread you can use the xG for a single chance as a metric to decide whether Azaz's chance was a sitter or not? Double standards.
I agree by the way that over the season xG will more or less level itself out like most stats and there'll be headers from that spot on the pitch that will get scored at 0.19 that will be much tougher than that in reality. But Azaz's chance was a LOT easier than a 1 in 5. It was a sitter.
BoroMart
Well-known member
Same
Laughing
Well-known member
Jesus, in no way shape or form did I use 1 example to say anything about azaz. I used a single stat, bad on hundreds of thousands of examples to demonstrate that what fans think of as a sitter is never as easy as people think. They are missed more often than they are scored.
If you don't understand the use of the single stat and why it was used, I can't help you understand.
Yes it's fine to dismiss xg over a single match. They can be misleading for lots of reasons a missed penalty being the most obvious one. I don't have a problem with that. However our xg over 9 games has been consistently good. Still a small sample, of course, but eventually a teams xg and the goals they score will converge.
BTW sofa score may have had azaz's chance at. 19 opta had it at 0.32. Should the goal have been scored more than 1 in 3? well it should be on target first and foremost, beyond that, I don't know.
Opta xg is based on hundreds of thousands of goal attempts. Whilst I don't necessarily agree with the score given for every chance, it is what it is. Going back to penalties I think a professional footballer should score 9+ out of ten not just below 8 but the stats say what they say. You can argue they are wrong but you don't have a way to measure that opinion so that's all it is, opinion.
Ravshoutsatbeck
Well-known member
The arguments are going around in circles again.
Have we had generally more opportunities to score than the opposition this season.
Yes.
Is that because of players becoming poor, having a poor run of form that will balance itself out with a poor run of form, or just bad luck?
That is the argument.
I do not really think it is luck.
Ipswich continued to be a statistical anomaly for the whole season last year, it happens. They barely missed chances and scored a load of worldies.
It might be that we continue to miss chances at an alarming rate despite having two players who mildly overachieved vs quality of chances last year.
A coach mostly sets up a team to create more chances than the opposition. The question is, could we be even more dominant, or could we create the same number of chances with players that were able to take more than expected?? That is for the manager to work out.
Have we had generally more opportunities to score than the opposition this season.
Yes.
Is that because of players becoming poor, having a poor run of form that will balance itself out with a poor run of form, or just bad luck?
That is the argument.
I do not really think it is luck.
Ipswich continued to be a statistical anomaly for the whole season last year, it happens. They barely missed chances and scored a load of worldies.
It might be that we continue to miss chances at an alarming rate despite having two players who mildly overachieved vs quality of chances last year.
A coach mostly sets up a team to create more chances than the opposition. The question is, could we be even more dominant, or could we create the same number of chances with players that were able to take more than expected?? That is for the manager to work out.
Redwurzel
Well-known member
From my memory of the game
Didn't their keeper save well from a decent effort from ELL and the rebound could have gone anywhere?- it went to Doak but not at a good angle and perfect pace for him to score with luck it would have been a fairly straight forward finish and the keeper would not have the chance to make a second save.
Other bits I remember:
Hamilton had a shot on goal if it deflected or the keeper was unseen it could have been a goal.
Their first goal benefited from Dieng been partially unsighted, he probably would have saved it if fully sighted or certainly got much more on it.
There was other decent chances too based on the stats (we had 5 attempts on goal to their 4, from official stats) but I can't recall them all.
What's the difference between a mis-hit and poor application? (to me they are the same for a Championship footballer i.e. a mis hit is poor application).
Teams and players will never score with every chance, no player is a perfect finisher there is always some degree of randomness and dry and wet spells of goal scoring. Watching a player like Bernie Slaven was like chalk and cheese and David Mills in the 1970s - sometimes they looked lower league and sometimes world beaters. ELL is starting to look like another player in their category.
IMO - Saturday's game was a draw based on stats and my observation of the overall game which I did attend.
Other Boro games where the result did not reflect the game certainly includes Pompey at home and Derby away. Both were like the Alamo.
Last season we were not worse than Rotherham or better than Leicester, but the results of the games said we were. A better indication was the final league table.
Redwurzel
Well-known member
Rav - are you sure about Ipswich? - when we played them at Portman Road they had 21 goal attempts, opposed to Watford's 9 attempts on Saturday and Ipswich only scored once. I usually equate about 9 attempts to 1 goal.
Where were Ipswich in the expected goals league table for the season - my guess is its certainly in the top 6.
edit - just found Ipswich expected goals for last season and it was plus 0.41 (expected goal scored per game take away expected goals conceded) this was the fourth highest in the division. Boro's was plus 0.13 the 6th highest.
Last edited:
Big_Nothing
Well-known member
And some stats' mothers are more useful than other stats' mothers…
I understand fully and you are using it incorrectly to make a point that is incorrect.
If you take the average speed of a hundred thousand vehicles down the M1. The average is 70MPH. The range is 50MPH to 148MPH. You are trying to tell us the car travelling 148MPH is actually traveling 70MPH because that was the average of hundreds of thousands of vehicles instead of looking at that one particular vehicle in isolation.
My point still stands. The xG rating of one in particular attempt is flimsy at best. It is a misuse of stats as you would say to use it to try and make a point about how difficult that one in particular attempt was.
As you know, as you have explained this multiple times, the xG doesn't take into account the height of the ball, whether he was marked and under pressure, whether he was balanced etc. The 0.19 rating is you rightly pointed out is determined from hundreds of thousands of headers taken from similar position and they will have wide range in difficulty from that same particular location. 19% are scored. The vast majority of these will not have been completely free headers where the ball is at the perfect height to head, both feet on the ground, full visibility of the cross with no obstruction or deflection and plenty of time to see it. I mean it is blatantly obvious to anyone who has any idea about football that that in particular chance was about as easy as a header gets from that location. ELL's header was from a very similar location so it stands to reason that it had a very similar xG (0.17). It is obviously a much more difficult chance. It's higher and he was being challenged. A 2% differential in apparent difficulty doesn't seem correct if you look at those two chances in isolation but I understand how it is calculated. The Azaz chance was a 145MPH Ferrari, the Lath one was probably your average 70MPH Ford Focus and there'll be some 50MPH HGVs along the season to come where the cross comes with no pace and is too high to really make a contact but as the striker makes contact with his head it will be deemed a 0.19 xG chance.
I don't know where you get your stats from, but Sofascore had it registered at 0.19, I like to use Sofascore as it gives the little map of where attempts were taken from and easier to identify which attempt is which.
FBRef also registers it at 0.19 and their stats are provided by Opta.
I have just checked Fotmob and they have it at 0.19, again it is provided by Opta.
But that is another issue if there are different xG ratings floating about for the same attempts. I assumed everywhere published Opta to be honest but it starts to make a total mockery if there are different providers using different formulas to determine xG ratings of the same attempt.
Unless you have just mis-read the data? Surely that couldn't be correct. Laughing making an error on reading the stats.
Last edited:
The chance is at 5:35. The keeper saves a rather tame effort from Lath and pushes it out into a bad area. If Doak had scored i think questions would have been asked of the keeper. Keeper redeems himself with a very good second save though. I mean we got a reasonable chance from the rebound, I don't think we can claim that it was unlucky that we never got a better chance. Had the keeper pushed it round the post or held it like he should have then it would never be mentioned that we were unlucky not to get a rebound.
This one is at 1:50.
It was probably worth an attempt but it wasn't even a half chance and it was a fairly comfortable save for the keeper. It would have taken a very good effort to beat the keeper from there.
Is this where we've got to? We need to hope for deflections?
I mean what are we determining a decent chance? I think we had 4 decent chances like I said in my first post.
Azaz header
Lath header
Doak rebound
The goal
I think we had our fair share of luck from the two bits of poor application from the opposition.
The mistake that led to the goal and the rebound that fell to Doak.
Apart from that we created 2 good chances which we squandered by missing the target.
Jonny_Rondos_Disco_pants
Well-known member
Im just sick of commentators and fans quoting stuff that is being pushed into the game to make it seem like they have some sort of edge over others.Tell me you are frightened of the intellectualising of football without telling me you are frightened of the intellectualisation of football
As a fan i’d rather just watch us win. The only stats that count at the end of a game are the goal tallys for both teams. Thats all fans and commentators need. I’d take a Karanka 1-0 win with 1 shot on goal over a 2-1 Carrick defeat with 30 shots on goal.
Coaches can use as much data as possible to fine tune players. Do fans really need to know about Km covered? Passing completion % ??? No.
festa5
Well-known member
They don't need to no. I find a lot of the stats interesting though as do many others.
It seems to me (and I'm not saying this applies to you by any stretch) that some fans don't like that the stats challenge their perception of what's actually happening.
It's quite an emotional game if you're truly invested in your team. I think it's pretty obvious many fans take defeat pretty badly and it's a natural reaction to want to blame someone for a defeat. The players, manager whoever. Its cathartic.
For many they will nearly always perceive us to have "played well" when we win and to have "played *****" when we lose, with very little nuance.
Yet football doesn't really work like that. It's a low scoring game and therefore often pretty unpredictable. The "better team" frequently doesn't win.
Often the stats directly contradict the perception of what some fans think they're seeing and they don't like it. Which is why I think so many get a bee in their bonnet whenever they're used to support arguments why we maybe weren't quite as bad in some games as some want to believe.
Laughing
Well-known member
I give up your average speed on the motorway example it's completely wrong. I mean it couldn't be more wrong when you compare it to xg.
It would be more accurate, if you want to use that analogy to measure average speeds at different times of the day and come up with an xS per time of day. You don't get the ridiculous range then.
Xg uses similar shots and averages them you have done something that is nothing like that.
My stats are paid for through optas api service. They are updated by a human reader after the event. Free versions may not do that. I can't explain what I don't know.
Finally. I can give you hundreds of examples of fans using the term sitter from our very own match day thread and you know this. Both azaz's chance and lath chance were described as sitters on Saturday by multiple members of this very forum. But you also knew that too.
You thought you had a gotcha, you didn't you misinterpreted what I said. This is why I get into arguments with people on stats subjects. I have every league game analysed since the mid 2020 season on every league above league 2 around Europe. Tens of thousands of games. I have applied hundreds of theories to those stats and tested those theories.
I learnt over time how to interpret them. It's not arrogance it's knowing my subject matter.
On the subject of stat providers. They are not all the same. I suspect Opta published the azaz mid at a similar 0.19 during the game where it's automated using ai. Opta them updated it after a human expert looked at it. It's a paid for service though.
However none of that matters. You think I used a single start to prove something about azaz. I didn't. I used a single stat that demonstrates something about how fans interpret a chance. They interpret chances very badly. My gran could score that, I could score that, it's a sitter, easier to score than miss and so on. But you know this, I am sure.
As I said if you still don't understand my reasoning after 2 attempts to explain it, I can't help you.
BoroPhil
Well-known member
Yep. A last minute goal turns an awful performance into a fantastic performance (or vice versa). Yet bar a moment at the end it is the same performance.
It's not wrong though is it. Not every attempt with the head from that spot on the pitch has the same chance of being scored. There is a range of difficulty due to the varying factors we've covered in my previous few posts. 0.19 (or 0.32) is the average within that range over hundreds of thousands of attempts. I am saying Azaz's chance will be at the very top end of that range due to the circumstances. The motorway analogy does work to describe this. I'm not using it as an analogy for xG as a whole but rather how that one particular xG rating is calculated. If Opta have adjusted the xG of that chance to a higher rating after the event like you suggest, then that only backs up my point.
Can you screenshot it? Or even better, can you screenshot all the xG attempts from that match to see how much they vary? Are you saying that a human will enter additional info after the event that affects the xG value? So it isn't 100% statistical? There is some human input that allows for bias or error?
It would also be interesting to see how the total xG for our players vary or the season as a whole rather than widely used values that are available on the sites mentioned. Fotmob, FBRef, Sofascore etc.
I still believe I am right regarding the range but you will never admit you have got it wrong. It's blatant hypocrisy but you are too arrogant to admit it.
Basically you can't be challenged on your interpretations, everyone else is wrong "because you know your subject matter".
Last edited:
Laughing
Well-known member
First point. Yes it is wrong. Your speed example is the equivalent of using an xg that rates every attempt the same. 1 in 10 or 0.1 BTW.
I'll screenshot Opta later. I can't show you the in game because that disappears once the chance had been regraded. Yes Opta look at chances and rescore them when they believe their algorithm got it wrong. Not sure of the criteria for that.
I don't record the game by game stats for individual players b g. In fact my statistical methods don't care, not should they, that's not how stats should be used. They should be used over as long a term as possible.
You are wrong about the range. I will say this again. I am not arrogant, I have spent the last 4 years analysing football stats. I know from experience the best way to interpret them often through throwing money away.
I have no problem when people disagree. Just expect to be challenged when I believe you are wrong. Challenging your perception doesn't mean I am right. The experience I have gained means I will, probably, be right more often than you. I will still be seeing plenty and have been see engel.
indeedido
Well-known member
Measures and Metrics need to be consistent in their definition and derivation. I'm sure given the focus, there will be ever more money poured into trying to make measures and metrics like XG more believable/convincing. In the meantime the source quoted should at least be the same one.
BG you are illustrating some points I made on the other thread really clearly.
According to Opta, Azaz had a combined XG of 0.375 against Watford and that weak header was the most likely to have been scored on average in the past, at 0.19. His other 3 efforts on goal yielded a combined XG of 0.18.
The header is one of his 3 misses classified as "Big Misses" in the season to date.
His other two were in the home matches against Swansea and Pompey and both deemed more likely to be goals with 0.3 and 0.25.
XG for a team may well tend to actual goals late in the season, but as with Ipswich last year (and I'm sure many others) it certainly won't always.
XG for a player may tend towards actual over a season, but more volatility here as there are far less incidents for a single player and more variation from the "average player".
But using XG to predict future performance is ridiculous (even laughing admits to having lost money pursuing this)
And using XG to assess performance to date is daft on a single match basis and little better over a small sample of matches.
I suspect the pros use XG and other manufactured stats only as a starting point, where questions of why a shot was unsuccessful begin, rather than taken as any sort of conclusion.
indeedido
Well-known member
Interesting final question Redwurzel.
The Actual Championship table for 2023-24. Hard Facts
Autos 1-2: Leicester 97, Ipswich 96 points
Play offs 3-6: Leeds 90, Soton 87, West Brom 75, Norwich 73
Relegated: Birmingham 50, Huddersfield 45, Rotherham 27
Championship based purely on Actual Goals Scored 23-24
Autos: Ipswich 92, Leicester 89
Play offs: Southampton 87, Leeds 81, Norwich 79, Boro 71
(West Brom 7th with 70).
Relegated: Millwall 45, Sheff Wed 44, Rotherham 37.
(Birmingham 18th with 50, Huddersfield 20th with 48)
Actual Goals Scored wasn't a bad proxy for points and actual position.
Championship based purely on Actual Goals Conceded 23-24
Autos: Leicester 41, Leeds 43
(Ipswich 7th with 57)
Play offs: West Brom 47, Bristol City 51, Sunderland 54, Millwall 55
(Southampton 14th with 63, Norwich 15th with 64)
Relegated: Blackburn 74, Huddersfield 77, Rotherham 89
( Birmingham 16th with 65)
Actual Goals Conceded decent proxy, but not as good as Goals Scored.
Championship based purely on Actual Goal Difference 23-24
Autos: Leicester +48, Leeds +38
(Ipswich 3rd +35)
Play offs: Ipswich +35, Southampton +24, West Brom +23, Norwich +15.
Relegated: Sheff Wed -24, Huddersfield -29, Rotherham -52.
(Birmingham 20th -15)
Actual Goal Difference was also a good proxy for points and actual position.
Championship based purely on XG 23-24
Autos: Southampton 76, Leicester 74.
(Ipswich 3rd 73)
Play offs: Ipswich 73, Leeds 71, Sunderland 68, Middlesbrough 66
(West Brom =14th 58, Norwich 10th 62)
Relegated: Sheff Wed/Cardiff/Millwall 54, Preston 49, Rotherham 45
(Birmingham =14th 58, Huddersfield 18th 57)
XG was poorly aligned to points and actual position.
Sunderland are a massive outlier given they finished only 16th
West Brom are massive outlier given they finished 6th
Preston are a big outlier given they finished 10th.
There are many big discrepencies in XG ranking versus actual league position.
What emerges is the 6 teams that went up or made play offs massively outscored their XG. Ipswich +19, West Brom +17, Leicester +15, Norwich +12, Southampton +11, Leeds +10
Yet, the 3 relegated teams are not the ones who must underscored their XG.
Sunderland -16, Stoke -11, QPR -11 were worst by some way.
Championship based purely on XGA 23-24
Autos: Leeds 46, =Leicester/West Brom 52
(Ipswich =5th 54)
Play offs: =Leicester/West Brom 52, Southampton 53, Ipswich 54, Sunderland 54
(Norwich 20th 67)
Relegated: Plymouth 69, Blackburn 70, Rotherham 78
(Birmingham 20th 67)
XGA was not strongly aligned to points and actual position.
Again Sunderland is a massive outlier finishing only 16th.
Norwich is an enormous outlier finishing 6th.
4 of the actual Top 6 conceded fewer than their XGA
Leicester -11, West Brom -5, Ipswich -3, Leeds -3, Norwich -3
Southampton conceded 10 more.
The 4 clubs who conceded far higher than XGA were
Huddersfield 12, Rotherham 11, Norwich 10, Sheff Wed 7
Championship based purely on XGD 23-24
Autos: Leeds 25, Southampton 23
(Leicester 3rd 22, Ipswich 4th 19).
Play offs Leicester 3rd 22, Ipswich 4th 19, Sunderland 5th 14, West Brom 6th 6.
(Norwich 14th -5)
Relegated Rotherham -33, Preston -17, Cardiff/Black/Plym/Birm -9
XGD got the promoted teams wrong.
Whilst Sunderland, Norwich and Preston are all massive outliers.
There are many others whose final position was very different to their position in the XGD rankings.
The XG, XGA and XGD do not form anywhere near a close match to actual final league positions.
There are many massive anomolies.
XG etc clearly doesn't match actual points and positions after 9 matches.
It didn't at the end of 23-24 Championship season.
They are flimsy indicators is all.
Good performing teams didn't always have high XG, but they always outperformed their XG.
Good performing teams didn't always have low XGA, but rarely did they not outperform it.
Which all really underlines the point that the XG measures are based on what all players have historically done as an average from approximate situations, not what good or bad players will do in specific circumstances.
It is very flawed.
It does perhaps indicate Sunderland fans look at X Tables rather than Actual tables and convince themselves they are real. According to XG measures for last season, Sunderland made the play offs 5th for XG, XGA and XGD. Yet I was sure they actually finished 16th.
This home made table summarises the measures and the numbers for 2023-24 Championship season.
It rolls across to the right
The X figures are from FootyStats site.
I can't find an X Points table for Championship 2023-24 to complete this little view - apologies.
| Club | Actual Points | Rank | Goals Scored | Rank | XG | Rank | Goals Against | Rank | XGA | Rank | Goal Difference | Rank | XGD | Rank | GS-XG | GA-XGA | GD-XGD | |
| Leicester | 97 | 1 | 89 | 2 | 74 | 2 | 41 | 1 | 52 | 2 | 48 | 1 | 22 | 3 | 15 | -11 | 26 | |
| Ipswich | 96 | 2 | 92 | 1 | 73 | 3 | 57 | 7 | 54 | 5 | 35 | 3 | 19 | 4 | 19 | 3 | 16 | |
| Leeds | 90 | 3 | 81 | 4 | 71 | 4 | 43 | 2 | 46 | 1 | 38 | 2 | 25 | 1 | 10 | -3 | 13 | |
| Southampton | 87 | 4 | 87 | 3 | 76 | 1 | 63 | 14 | 53 | 4 | 24 | 4 | 23 | 2 | 11 | 10 | 1 | |
| West Brom | 75 | 5 | 70 | 7 | 58 | 14 | 47 | 3 | 52 | 2 | 23 | 5 | 6 | 6 | 12 | -5 | 17 | |
| Norwich | 73 | 6 | 79 | 5 | 62 | 10 | 64 | 15 | 67 | 20 | 15 | 6 | -5 | 14 | 17 | -3 | 20 | |
| Hull | 70 | 7 | 68 | 9 | 65 | 7 | 60 | 10 | 61 | 11 | 8 | 9 | 4 | 8 | 3 | -1 | 4 | |
| Middlesbrough | 69 | 8 | 71 | 6 | 66 | 6 | 62 | 13 | 60 | 9 | 9 | 8 | 6 | 6 | 5 | 2 | 3 | |
| Coventry | 64 | 9 | 70 | 7 | 63 | 8 | 59 | 9 | 63 | 15 | 11 | 7 | 0 | 11 | 7 | -4 | 11 | |
| Preston | 63 | 10 | 56 | 14 | 49 | 23 | 67 | 18 | 66 | 19 | -11 | 15 | -17 | 23 | 7 | 1 | 6 | |
| Bristol City | 62 | 11 | 53 | 15 | 57 | 18 | 51 | 4 | 60 | 9 | 2 | 10 | -3 | 12 | -4 | -9 | 5 | |
| Cardiff | 62 | 12 | 53 | 15 | 54 | 20 | 70 | 20 | 63 | 15 | -17 | 21 | -9 | 19 | -1 | 7 | -8 | |
| Millwall | 59 | 13 | 45 | 22 | 54 | 20 | 55 | 6 | 59 | 7 | -10 | 14 | -5 | 14 | -9 | -4 | -5 | |
| Swansea | 57 | 14 | 59 | 12 | 58 | 14 | 65 | 16 | 63 | 15 | -6 | 13 | -5 | 14 | 1 | 2 | -1 | |
| Watford | 56 | 15 | 61 | 10 | 63 | 8 | 61 | 12 | 62 | 13 | 0 | 11 | 1 | 9 | -2 | -1 | -1 | |
| Sunderland | 56 | 16 | 52 | 17 | 68 | 5 | 54 | 5 | 54 | 5 | -2 | 12 | 14 | 5 | -16 | 0 | -16 | |
| Stoke | 56 | 17 | 49 | 19 | 60 | 12 | 60 | 10 | 59 | 7 | -11 | 15 | 1 | 9 | -11 | 1 | -12 | |
| QPR | 56 | 18 | 47 | 21 | 58 | 14 | 58 | 8 | 62 | 13 | -11 | 15 | -4 | 13 | -11 | -4 | -7 | |
| Blackburn | 53 | 19 | 60 | 11 | 61 | 11 | 74 | 22 | 70 | 23 | -14 | 19 | -9 | 19 | -1 | 4 | -5 | |
| Sheff Wed | 53 | 20 | 44 | 23 | 54 | 20 | 68 | 19 | 61 | 11 | -24 | 22 | -7 | 17 | -10 | 7 | -17 | |
| Plymouth | 51 | 21 | 59 | 12 | 60 | 12 | 70 | 20 | 69 | 22 | -11 | 15 | -9 | 19 | -1 | 1 | -2 | |
| Birmingham | 50 | 22 | 50 | 18 | 58 | 14 | 65 | 16 | 67 | 20 | -15 | 20 | -9 | 19 | -8 | -2 | -6 | |
| Huddersfield | 45 | 23 | 48 | 20 | 57 | 18 | 77 | 23 | 65 | 18 | -29 | 23 | -8 | 18 | -9 | 12 | -21 | |
| Rotherham | 27 | 24 | 37 | 24 | 45 | 24 | 89 | 24 | 78 | 24 | -52 | 24 | -33 | 24 | -8 | 11 | -19 |