I appreciate some people only care about one stat and that’s the actual result of the game, the points on the board and the league position. But I love a stat. @reids (or anyone else) would it be possible to update the expected goals for and against league table in here each week? Plus any other interesting (for people like me) stats in here throughout the season? Those who hate all this stuff can obviously not return, or pop in to mock stat lovers like myself! Thanks in advance!
A bit like we had last season but it sort of got forgotten (not surprising considering how shyte it was). I'll pin this thread - so @reids or whoever wants to can update after each game.
Will post the stats from each match in here rather than in the match day threads, along with the xG difference league table each week. - Coventry won the xG battle, although a good chunk of that was from the penalty (0.78 xG) - Rajovics first goal was 0.41 xG and his 2nd was 0.42 xG. Martins goal was only 0.05 xG (0.55 PSxG) - Their FK goal was only 0.03 xG Top of the xG Difference table (that's the real quiz) although it's missing the WBA/Huddersfield + Norwich/Rotherham games. We've conceded the least amount of xG in the league (3.48 xG) which backs up that we're causing our own issues by being sloppy/poor GKing.
Love this, thank you @reids plus an extra doff of the cap for The Office quote which made me chuckle. We look in good company amongst that top 8, looks like VI system is working well, hopefully the players can put a couple more away and cut out a few of the errors and it could actually be a decent season!
Pinched a win in a tight game. - Our lowest xG score this season. - Rajovics goal was only 0.05 xG and Andrews goal was only 0.03 xG. - Our "best" chance of the game (aka most xG) was from Porteous in the 78th min (0.09 xG). - With 0.83 xG they also didn't create much - especially since their 2 best chances (0.37 xG + 0.21 xG - Bachmanns double save) came from the same passage of play.
Thanks @reids , we’ve had games with the better XG and not won and now the opposite. Funny old game eh Saint
Thanks Reids, great stats as always. I'm interested to know though why the model for the two Birmingham chances in quick succession adds the xG together, when they were clearly not independent events. To give an extreme example, if a team had a penalty (0.75 xG) which was saved, but then presented an easy tap-in on the rebound (e.g. another 0.75 xG), it would seem odd to say that the penalty was worth 1.5 xG! Whereas to me it would seem more logical that a fair xG for that passage of play would be less than 1 but greater than 0.75, as the second attempt only happened because the first attempt was saved.
Yeah I fully agree, whilst it's useful to have the xG values of both shots (to rate keeper + attacker) it does seem illogical to include both in the final xG values. There doesn't seem to be a definition of "cumulative xG" that's in the match stats for each side but that could be a way of adjusting it (since that is lower than their overall xG)?
Also the second Bachmann save we had 2-3 players blocking the ball behind him and I think XG only accounts for defenders between the shooter and the keeper not behind. .
@EB Hornet, you asked elsewhere if I had any stats up my sleeve, so to speak. So, for you, and any other stats fans here is a ranking of all current league clubs. I have given each a score, based on the tiers they have played in over the last 30 seasons (including the current one), so from the third Premier League season onwards. I have weighted the tiers, 16,8,4,2,1 so the maximum possible score is 480. I have then expressed these scores as a percentage of 480. I am pleased to see that we are ranked 24th, comfortably in the top 30 (a la GT's hope) and that our lovely neighbours do not make the top 50.
Tight game with some good goals. - Martins goal was worth 0.04 xG and Inces was worth 0.07 xG - Kones first half 1 vs 1 was our best chance of the match, worth 0.21 xG. His header in the 2nd half was our 2nd best chance, worth 0.16 xG - Their FK goal was worth 0.08 xG and their 2nd was worth 0.14 xG. - Bachmanns save against Thomas-Asante in the 79th min was their best chance - worth 0.30 xG (42% of their total!)
Nice work, like the concept / idea you’ve come up with. Good to see we are that high, and in good company there. The Pozzo era must have risen us up this table - although if you did the 30 years up to the Pozzo era we maybe were similar as this would include GT’s first top flight era?
Thanks @reids i really like this stat analysis after the games. I’m surprised our XG wasn’t a bit higher but I guess our 2 goals were great strikes. Very pleasing how few chances we give away, which seems to go against everyone’s fears about our style of play. From what I’m seeing (more so at home) we are keeping possession and regaining it pretty quickly. On that last point, is there a stat available for how quickly each team recovers possession? It seems to me we are doing this very quickly this season.
Yeah there's a few, Passes/Defensive Action (also known as PPDA) is probably the best for measuring that. This stat basically counts how many passes a team allows before making a defensive action (tackle, interception, foul), meaning a team with a low PPDA allows less passes before they manage to disrupt the opponent (basically they're successful in pressing more). As you can see, you're correct and we have the 3rd lowest PPDA score in the league (a good thing!)
Awesome! Thanks for the stats and the explanation. I’ve very much enjoyed this part of our game this season.
Tough day out for the Hornets. - Our best chance was Bayos shot from Asprillas cross that registered 0.15 xG (guessing the height of the ball impacted the xG total there) - Our lowest xG total so far this season - Lowest amount of possession we've had a single match so far this season and our lowest pass completion %, Leeds were very good at disrupting the spells when we did have the ball. - Bachmanns save against Rutter in the 6th minute was the biggest chance of the game (0.42 xG) - Goals: Piroes opener was 0.40 xG, Byrams header was 0.09 and Anthonys 1 vs 1 was 0.32 xG
Stats probably a bit misleading since they scored 2 early goals and likely had little impetus to attack but still: - We created 1.84 xG with Bayo responsible for 70% of our total (1.28 xG!) - Bayos goal was worth 0.36 xG, his big miss after rounding the keeper was worth 0.62 xG. Hoedts goal was worth 0.04 xG - Their first goal was worth 0.16 xG with the 2nd and 3rd goals being worth 0.13 xG and 0.20 xG. - Their 3rd goal was their biggest chance of the game (0.20 xG), with McGrees blocked shot after a scramble from a corner worth 0.19 xG - As mentioned pre-game, their game flows through Hackney and that was reflected through how much he got the ball in the first half.
Appreciate I'm replying to this post very late but shows limitations of using xG for single incidents rather than aggregating over games/seasons I think. There's no way that Rajovic's first goal against Coventry (on the stretch volley from a cross fizzed in, keeper to beat) is almost identical likelihood of a goal as his second (keeper taken out of the game by Bayo, open net).
xG definitely has it's limitations in single incidents and even over the course of a single game also - but still interesting to see. I think both would've been much higher without the close proximity of the #22 which would've lowered the totals. Both goals go up to 0.99 when looking at PSxG though.
What's PSxG? Screenshots above make it look a lot closer than I thought it was. I have therefore decided to scrap my model which is based on memory of events over two weeks ago!
PSxG is basically what happens after the shot is taken and is essentially the xG value of shots if they were on target (you don't get a PSxG value for shots that were missed). So for instance looking at Swifts FK goal against us the other week - the initial shot is 0.08 xG (aka location, players in the way etc) but the PSxG is 0.49. So provided both of Rajs goals against Cov were on target, considering the positions of the defence + keeper they're basically a near certainty to scoring. You can use xG + PSxG in combination to evaluate a players shooting ability, if their PSxG exceeds their xG it could be a good sign that they're a good finisher
Another frustrating game then - although we knew that without the stats - but this definitely backs it up. Thanks as always Reids
A game with enough chances for us to take the 3 points - Their best chance was their goal (0.32 xG) - Set-pieces made up 76% of their xG - Bayos goal was worth a whopping 0.92 xG - We missed a few big chances - Ince blazing over was worth 0.38 xG. Asprillas header was worth 0.66 xG
I would have put the other way round as the cross was just too high for Asprilla (do they take height of player in to account?) and Ince should have scored
They don't take the height of the player or aerial ability into account (but does take into account the balls height), Asprillas was likely so high as there was nobody between him + goal and the close proximity to goal. Reckon Inces was lower as was further out and the pressure from the defender probably knocked it down a bit as well
Interesting graphic from Opta on their socials this morning, looking at the expected points (purple) vs actual points (red) in the league and as backed up by the eye test - we're one of the teams with the biggest differences between expected + actual points
As we all know, it was a game devoid of quality: - Best chances of the match were headers - Hendricks header from a free kick and a first half header from Louza (both 0.12 xG) - Asprillas goal was 0.05 xG (and 0.65 PSxG) - It was our 4th worst game in terms of xG creation (only Sunderland, Leeds + Birmingham were worse) - It was also our 3rd best game in terms of xG conceded (only QPR + Blackburn created less)
Again a game of few chances for either team - Bayos header was the best chance of the game (0.29 xG) - Semas goal was worth just 0.03 xG and 0.05 PSxG - Their best chance was Darlings header from a freekick in the first 6 mins of the game (0.20 xG) Any requests for stats, visualisations, trendlines of metrics etc please let me know!
Surprised their early chance was that low in terms of xG, he made a bit of a hash of it, but it was a good chance from around the 6 yard area
I was surprised by this as well as thought it would've been higher considering the quality of the strike. Watching it back the GK is very very well positioned and almost gets a hand to it I think 1 in 5 is fair enough - despite there being nobody in the way (other than Bachmann) headers are harder to score which will impact the xG value, guessing these sort of chances are more common than we think!