| Opta Analyst 10:26 - Dec 12 with 1912 views | dirtyboy |
Lovely bit data to pour over this morning having had it all been updated.
https://theanalyst.com/eu/2023/12/is-the-championship-table-lying-to-us/
Suggests on underlying numbers, we should be 4th on 35 points (still would be incredible) and we know McKenna does place credence on numbers...given we were top of this metric last year, we knew we were the best team 'on paper' so it's not to be sniffed at, but doesn't work out perfectly.
Does suggest we can place some trust in it.
Transfer those numbers to the rest of the season and Leeds would fall short by 3 points to catch us in second.
However...clearly doesn't work like that!
It's going to be an incredible season whatever way we look at it!
Leeds top [Post edited 12 Dec 2023 10:27]
|  | | |  |
| Opta Analyst on 10:55 - Dec 12 with 1761 views | GavTWTD |
Expected goals is still a bit of a mystery to me. I'm assuming there is no one method of calculating XG. |  |
| |
| Opta Analyst on 10:58 - Dec 12 with 1752 views | glasso |
| Opta Analyst on 10:55 - Dec 12 by GavTWTD |
Expected goals is still a bit of a mystery to me. I'm assuming there is no one method of calculating XG. |
Leeds fans expect to score 10 every game. I think that's how it works. |  | | |
| Opta Analyst on 11:04 - Dec 12 with 1689 views | FrankfurtBlue |
I must admit to not having spent much time trying to understand xG etc, but i have two observations.
Firstly, there must be some subjectivity in assessing a goal chance because, put simply, someone has to decide how easy or difficult a goal scoring chance is and then it has to be compared to all past chances put into that category.
Secondly, as far as I know the metric doesn't take into account the relative abilities of the players involved, e.g. the Marcus Harness goal v Brum would get the same xG rating whether he takes a swing at it or, ahem with the greatest of respect, Cole Skuse does. Likewise, is the keeper's ability taken into account?
There are probably more reasons to discount this "analysis" either completely or greatly. |  | | |
| Opta Analyst on 11:17 - Dec 12 with 1620 views | SheffordBlue |
| Opta Analyst on 11:04 - Dec 12 by FrankfurtBlue |
I must admit to not having spent much time trying to understand xG etc, but i have two observations.
Firstly, there must be some subjectivity in assessing a goal chance because, put simply, someone has to decide how easy or difficult a goal scoring chance is and then it has to be compared to all past chances put into that category.
Secondly, as far as I know the metric doesn't take into account the relative abilities of the players involved, e.g. the Marcus Harness goal v Brum would get the same xG rating whether he takes a swing at it or, ahem with the greatest of respect, Cole Skuse does. Likewise, is the keeper's ability taken into account?
There are probably more reasons to discount this "analysis" either completely or greatly. |
On your first point - it's based purely on how many times a goal was scored from a particular position as a %. Their models run hundreds of thousands of shots through their system and identify which ones went in from where. Most models now add in the positions of opposition players and some add additional levels of complexity. That will inevitably miss out some factors.
On your second point - the players abilities are not considered but that's viewed as a strength of the modelling rather than a flaw. The question is 'how likely is it a player' will score from there rather than 'how likely is it that player x will score' from there.
xG is far from perfect but I think it tends to give a better sense of whether teams are scoring from high quality chances or relying on worldies.
This article does note that game state isn't considered in their overall analysis which for me is one of the reasons that we are viewed as 'overperforming'. In reality while we've edged some close games (and in the case of the Leeds game didn't edge it) there have been very few where on the balance of play we didn't deserve the points we got. | |
| |
| Opta Analyst on 11:22 - Dec 12 with 1572 views | HighgateBlue |
At the other end of the table, this suggests that Wednesday might even have it in them to stay up (seemingly having been wildly unlucky so far), and Plymouth may well be in danger of a relegation battle if they're not careful. |  | | |
| Opta Analyst on 11:39 - Dec 12 with 1497 views | FrankfurtBlue |
| Opta Analyst on 11:17 - Dec 12 by SheffordBlue |
On your first point - it's based purely on how many times a goal was scored from a particular position as a %. Their models run hundreds of thousands of shots through their system and identify which ones went in from where. Most models now add in the positions of opposition players and some add additional levels of complexity. That will inevitably miss out some factors.
On your second point - the players abilities are not considered but that's viewed as a strength of the modelling rather than a flaw. The question is 'how likely is it a player' will score from there rather than 'how likely is it that player x will score' from there.
xG is far from perfect but I think it tends to give a better sense of whether teams are scoring from high quality chances or relying on worldies.
This article does note that game state isn't considered in their overall analysis which for me is one of the reasons that we are viewed as 'overperforming'. In reality while we've edged some close games (and in the case of the Leeds game didn't edge it) there have been very few where on the balance of play we didn't deserve the points we got. |
Thanks for that info.
It sounds very static as a measure. Do you know whether, for example, the speed and angle with which the ball arrives for the scoring chance is taken into account? There are so many factors that make it no more than a rough indicator of chance conversion, but certainly not a reliable tool for predicting league positions.
I agree with your last point....I cannot think of a game this season where we clearly came away with stolen points. As KMc says, "fine margins"......away at QPR for example. | | | |
| Opta Analyst on 11:44 - Dec 12 with 1475 views | blueasfook |
Maths nerds need to stay out of football.
The ONLY thing that tells the real story is the table, and we are currently second. |  |
| |
| Opta Analyst on 11:45 - Dec 12 with 1454 views | _clive_baker_ |
| Opta Analyst on 11:17 - Dec 12 by SheffordBlue |
On your first point - it's based purely on how many times a goal was scored from a particular position as a %. Their models run hundreds of thousands of shots through their system and identify which ones went in from where. Most models now add in the positions of opposition players and some add additional levels of complexity. That will inevitably miss out some factors.
On your second point - the players abilities are not considered but that's viewed as a strength of the modelling rather than a flaw. The question is 'how likely is it a player' will score from there rather than 'how likely is it that player x will score' from there.
xG is far from perfect but I think it tends to give a better sense of whether teams are scoring from high quality chances or relying on worldies.
This article does note that game state isn't considered in their overall analysis which for me is one of the reasons that we are viewed as 'overperforming'. In reality while we've edged some close games (and in the case of the Leeds game didn't edge it) there have been very few where on the balance of play we didn't deserve the points we got. |
Looking back on our last 13 games, I think you can argue they've all been reasonable results that have reflected the games. 2 deserved defeats to WBA and Fulham, 3 draws at Rotherham, Brum and Huddersfield which were probably about right on balance, and 8 wins. If anything, I think a couple of those score lines don't do justice to our dominance such as Millwall, Swansea and Coventry.
I think there's some merit that we came out on the right side of some fine lines earlier in the season though, I think back to Cardiff where we're 2-0 down and perhaps a little fortuitous with the winner, Sunderland away and QPR away could've been draws on another day. I think it's a stretch to suggest we're 12.5 points better off IMO, but perhaps low 40's rather than high 40's as a points tally wouldn't be unreasonable when just applying my eyeball test.
We're not though, we've got the points we've got, and 2.4 ppg is absolutely insane. Whoever finishes in the top 2 this season will have earned it and be there on merit. |  | | |
Login to get fewer ads
| Opta Analyst on 12:49 - Dec 12 with 1301 views | itfc_bucks |
| Opta Analyst on 11:17 - Dec 12 by SheffordBlue |
On your first point - it's based purely on how many times a goal was scored from a particular position as a %. Their models run hundreds of thousands of shots through their system and identify which ones went in from where. Most models now add in the positions of opposition players and some add additional levels of complexity. That will inevitably miss out some factors.
On your second point - the players abilities are not considered but that's viewed as a strength of the modelling rather than a flaw. The question is 'how likely is it a player' will score from there rather than 'how likely is it that player x will score' from there.
xG is far from perfect but I think it tends to give a better sense of whether teams are scoring from high quality chances or relying on worldies.
This article does note that game state isn't considered in their overall analysis which for me is one of the reasons that we are viewed as 'overperforming'. In reality while we've edged some close games (and in the case of the Leeds game didn't edge it) there have been very few where on the balance of play we didn't deserve the points we got. |
I get that there are models for this sort of thing, but there must (should?) be some accountability for *who* is taking the shot.
Picture the scene - corner is whipped in, cleared by a defender, falls to a midfielder on the edge of the box, he draws back to fire it back in....the model says there's an xG of whatever for this, but we all know we'd rather have Broadhead taking on that shot than, for arguments sake, Cole Skuse. | | | |
| Opta Analyst on 12:55 - Dec 12 with 1268 views | heavyweight |
Perhaps this just means that we are a bit more circumspect before an attempt at goal. Eg not taking a shot when it is likely to be blocked or hail maryvatempts from outside the area.
From remembering quite a few of our goals have come from breakaways and in those instances the likely hood of a goal based on where the shot was taken from is going to be far higher than if it was from general play. Not sure this is taken into account in XG.
Also Chaplin has been exceptionally clinical this season. | | | |
| Opta Analyst on 13:01 - Dec 12 with 1235 views | ChorleyBoy |
What I take from this is that Opta is convinced like everyone outside Norwich that the Budgies are a bottom half of the Championship team. |  | | |
| Opta Analyst on 13:28 - Dec 12 with 1136 views | SheffordBlue |
| Opta Analyst on 12:49 - Dec 12 by itfc_bucks |
I get that there are models for this sort of thing, but there must (should?) be some accountability for *who* is taking the shot.
Picture the scene - corner is whipped in, cleared by a defender, falls to a midfielder on the edge of the box, he draws back to fire it back in....the model says there's an xG of whatever for this, but we all know we'd rather have Broadhead taking on that shot than, for arguments sake, Cole Skuse. |
I think for this and the other point that Frankfurt Blue raises about the speed the ball is received at etc that the model uses enough reference points that it balances out. E.g If 1000 shots have been taken from a certain point in the penalty box then x% will have been struck by forwards, x% by midfielders, etc.
As a stat/model it's less useful for predictions than telling us what happened historically. e.g In the early part of last season Plymouth were relying on low % (low xG) chances for quite a few of their wins. Schumacher made some changes so that later on they were less reliant on this. | |
| |
| Opta Analyst on 13:34 - Dec 12 with 1094 views | DJR |
| Opta Analyst on 12:49 - Dec 12 by itfc_bucks |
I get that there are models for this sort of thing, but there must (should?) be some accountability for *who* is taking the shot.
Picture the scene - corner is whipped in, cleared by a defender, falls to a midfielder on the edge of the box, he draws back to fire it back in....the model says there's an xG of whatever for this, but we all know we'd rather have Broadhead taking on that shot than, for arguments sake, Cole Skuse. |
That was a point I was going to make. |  | | |
| Opta Analyst on 13:36 - Dec 12 with 1078 views | blueasfook |
| Opta Analyst on 13:01 - Dec 12 by ChorleyBoy |
What I take from this is that Opta is convinced like everyone outside Norwich that the Budgies are a bottom half of the Championship team. |
I liked in the charts that they were the only team that are neither under-achieving or over-achieving, but right where they should be. | |
| |
| Opta Analyst on 13:42 - Dec 12 with 1057 views | Herbivore |
| Opta Analyst on 11:17 - Dec 12 by SheffordBlue |
On your first point - it's based purely on how many times a goal was scored from a particular position as a %. Their models run hundreds of thousands of shots through their system and identify which ones went in from where. Most models now add in the positions of opposition players and some add additional levels of complexity. That will inevitably miss out some factors.
On your second point - the players abilities are not considered but that's viewed as a strength of the modelling rather than a flaw. The question is 'how likely is it a player' will score from there rather than 'how likely is it that player x will score' from there.
xG is far from perfect but I think it tends to give a better sense of whether teams are scoring from high quality chances or relying on worldies.
This article does note that game state isn't considered in their overall analysis which for me is one of the reasons that we are viewed as 'overperforming'. In reality while we've edged some close games (and in the case of the Leeds game didn't edge it) there have been very few where on the balance of play we didn't deserve the points we got. |
Over the years I've come round to the view that xG is useful as a metric but is also flawed. I do think our outperforming our xG for isn't down to luck, it's down to us creating chances for our most clinical finishers and often doing so in ways that increase the probability of us scoring. Take the Chaplin goal against Hull, for example. It's probably quite a low xG chance from 20 yards in the centre of the goal and many strikes from there are probably first time or somewhat rushed. But the way Chaplin was set up meant he could take a touch and pick his spot, and with the quality he has on his left foot, I think most of the ground were expecting the net to rustle there. Now, that's not very scientific but when you factor in other variables, what was on paper a low xG chance probably wasn't. We've probably had a few others that are similar in that respect, we're good at picking out good finishers in space in decent areas. That said, as a general measure of how creative teams are I do think xG has value. [Post edited 12 Dec 2023 17:05]
|  |
| |
| Opta Analyst on 16:57 - Dec 12 with 837 views | GeoffSentence |
| Opta Analyst on 13:36 - Dec 12 by blueasfook |
I liked in the charts that they were the only team that are neither under-achieving or over-achieving, but right where they should be. |
Won't be long before they are pining for a 15th place finish. |  |
| |
| |
By continuing to use the site, you agree to our use of cookies and to abide by our Terms and Conditions.
