The only slightly more advanced than shots/possession statistics thread

URZZZZ
Hob Nob Addict
Posts: 7279
Joined: 20 Apr 2013 18:30

Re: The only slightly more advanced than shots/possession statistics thread

by URZZZZ » 11 Apr 2019 13:55

Strong home record, they haven't lost a home game in 2019 apart from the 1st January possibly? Considering they've still got 3 home games. But the odds are quite high considering their distance from the drop

Would still put it at Wigan>Reading>Rotherham>Millwall to go down but who knows. Maybe QPR but with Blackburn and Forest at home still possibly the two easiest games left of the season, think they'll be fine

Hound
Hob Nob Legend
Posts: 24934
Joined: 27 Sep 2016 22:16
Location: Simpleton

Re: The only slightly more advanced than shots/possession statistics thread

by Hound » 11 Apr 2019 14:06

Wigan have 4 points from hone games with Ipswich, Brentford, Boro and Stoke, with the win against Bolton in there as well

So 7 points from 5 very winnable games, including 2 against absolute dross

Wouldn’t call that particularly strong

Brain Traysers
Member
Posts: 663
Joined: 06 Jul 2004 12:02
Location: Cascadia

Re: The only slightly more advanced than shots/possession statistics thread

by Brain Traysers » 11 Apr 2019 17:24

RG30 Fivethirtyeight had Rotherham 41% chance relegation and us at 39% chance prior to last night's fixtures. Now gone 54% Rotherham and 22% ourselves. Still got a SPI of a bottom 3 side :lol:


SPI is a slow moving measure so that will persist for the rest of the season, pretty much regardless of our results. In addition, if we stay up, that model will likely have us favourite, or second favourite to go down (Luton and Sunderland are already in the mid 40s, half the teams in play off positions have a higher score than us). Its not a coincidence that 538 and my xG ==> xP bits both put us as overachieving as the ultimate source is the same data.

Ben Mayhew (@Experimental361) also now has us at around 21%, with Rotherham at about 62%. Those numbers were 36% and 48% respectively before this round of games - a big jump for only 1pt difference.


Brain Traysers
Member
Posts: 663
Joined: 06 Jul 2004 12:02
Location: Cascadia

Re: The only slightly more advanced than shots/possession statistics thread

by Brain Traysers » 11 Apr 2019 17:52

Away from relegation probabilities, Ram Srinivas (@ramesss) produced a couple of interesting charts on Championship fullbacks

Striking that our starting two are almost identical, and Gunter only differs by being 4pp lower for passing accuracy. All 3 are incredibly consistent in only playing about 12% of their passes in the final third (vs 15.5% league avg)- this strikes me as quite different to the popular narrative of Yiadom and his surging forward runs as the evidence doesn't support him playing more passes in advanced areas.


Again our three full backs look very similar - with 16-20% of crosses being low (bang on the divisional average). I was surprised to see Yiadom (1.9/90) sat between Gunter (2.5/90) and Blackett (1.3/90) for crosses/90.

Both seems to offer some tentative evidence of a halo effect on (POTS candidate) Yiadom, presumably from him (i) being a new signing and (ii) for taking Gunter's first team place. Of course the missing and key piece is the defensive side of their games, where Yiadom might come out way in front of the others, but I'm not convinced (at a glance of whoscored it's much of a muchness).

The final thing that stood out was just how major an attacking threat Obita can be for us when fit - he averages 5.5 crosses/90 over his career, higher than anyone on the second chart based of this season (but I doubt he would have hit numbers quite so high with our current team).

muirinho
Hob Nob Regular
Posts: 2075
Joined: 20 Jan 2016 12:10

Re: The only slightly more advanced than shots/possession statistics thread

by muirinho » 11 Apr 2019 18:47

Brain Traysers ....
Striking that our starting two are almost identical, and Gunter only differs by being 4pp lower for passing accuracy. All 3 are incredibly consistent in only playing about 12% of their passes in the final third (vs 15.5% league avg)- this strikes me as quite different to the popular narrative of Yiadom and his surging forward runs as the evidence doesn't support him playing more passes in advanced areas.

....
Again our three full backs look very similar - with 16-20% of crosses being low (bang on the divisional average). I was surprised to see Yiadom (1.9/90) sat between Gunter (2.5/90) and Blackett (1.3/90) for crosses/90.

Both seems to offer some tentative evidence of a halo effect on (POTS candidate) Yiadom, presumably from him (i) being a new signing and (ii) for taking Gunter's first team place. Of course the missing and key piece is the defensive side of their games, where Yiadom might come out way in front of the others, but I'm not convinced (at a glance of whoscored it's much of a muchness).

The final thing that stood out was just how major an attacking threat Obita can be for us when fit - he averages 5.5 crosses/90 over his career, higher than anyone on the second chart based of this season (but I doubt he would have hit numbers quite so high with our current team).


In 2013/2014 Obita won POTS in his first full season in the first team. Gunter was 2nd (2nd season) and Williams 3rd (1st season).

I think all players, unless they are particularly apalling or obnoxious, get a bit of a halo effect when they join first - because we are excited by what they can bring. But, as time wears on, we begin to forget what they can do, and are more aware of what they can't. Their faults become more prominent and irritate us more.

A bit like a relationship, all lovey-dovey to begin with, but after a couple of years, it's the teabags left in the sink, and clothes not put away that come to the fore.

And maybe a few years down the line, when you're with someone else, you start forgetting about the teabags, but remember the breakfasts in bed. The halo returns when you start comparing their good points to the faults of the one that's around right now.

So, as a general rule, if new players and players from long ago seem that much better than players that have been here a few years, at least part of that is going to be "new halo, reverse halo, renewed halo" effect.

Having said all that, probably still going to go for one of Martinez, Rino and Yids as POTS!


User avatar
Snowflake Royal
Hob Nob Legend
Posts: 39401
Joined: 20 Jun 2017 17:51

Re: The only slightly more advanced than shots/possession statistics thread

by Snowflake Royal » 11 Apr 2019 19:19

Brain Traysers Away from relegation probabilities, Ram Srinivas (@ramesss) produced a couple of interesting charts on Championship fullbacks

Striking that our starting two are almost identical, and Gunter only differs by being 4pp lower for passing accuracy. All 3 are incredibly consistent in only playing about 12% of their passes in the final third (vs 15.5% league avg)- this strikes me as quite different to the popular narrative of Yiadom and his surging forward runs as the evidence doesn't support him playing more passes in advanced areas.


s'cos he doesn't pass half the time he's in the final third, he loses it or shoots. :wink:

User avatar
leon
Hob Nob Legend
Posts: 29043
Joined: 14 Apr 2004 09:18
Location: Hips, Lips, Tits, Power

Re: The only slightly more advanced than shots/possession statistics thread

by leon » 11 Apr 2019 20:24

muirinho
Brain Traysers ....
Striking that our starting two are almost identical, and Gunter only differs by being 4pp lower for passing accuracy. All 3 are incredibly consistent in only playing about 12% of their passes in the final third (vs 15.5% league avg)- this strikes me as quite different to the popular narrative of Yiadom and his surging forward runs as the evidence doesn't support him playing more passes in advanced areas.

....
Again our three full backs look very similar - with 16-20% of crosses being low (bang on the divisional average). I was surprised to see Yiadom (1.9/90) sat between Gunter (2.5/90) and Blackett (1.3/90) for crosses/90.

Both seems to offer some tentative evidence of a halo effect on (POTS candidate) Yiadom, presumably from him (i) being a new signing and (ii) for taking Gunter's first team place. Of course the missing and key piece is the defensive side of their games, where Yiadom might come out way in front of the others, but I'm not convinced (at a glance of whoscored it's much of a muchness).

The final thing that stood out was just how major an attacking threat Obita can be for us when fit - he averages 5.5 crosses/90 over his career, higher than anyone on the second chart based of this season (but I doubt he would have hit numbers quite so high with our current team).


In 2013/2014 Obita won POTS in his first full season in the first team. Gunter was 2nd (2nd season) and Williams 3rd (1st season).

I think all players, unless they are particularly apalling or obnoxious, get a bit of a halo effect when they join first - because we are excited by what they can bring. But, as time wears on, we begin to forget what they can do, and are more aware of what they can't. Their faults become more prominent and irritate us more.

A bit like a relationship, all lovey-dovey to begin with, but after a couple of years, it's the teabags left in the sink, and clothes not put away that come to the fore.

And maybe a few years down the line, when you're with someone else, you start forgetting about the teabags, but remember the breakfasts in bed. The halo returns when you start comparing their good points to the faults of the one that's around right now.

So, as a general rule, if new players and players from long ago seem that much better than players that have been here a few years, at least part of that is going to be "new halo, reverse halo, renewed halo" effect.

Having said all that, probably still going to go for one of Martinez, Rino and Yids as POTS!


I never realisedyou’ were a woman...

Brain Traysers
Member
Posts: 663
Joined: 06 Jul 2004 12:02
Location: Cascadia

Re: The only slightly more advanced than shots/possession statistics thread

by Brain Traysers » 27 Apr 2019 01:03

Ram Srinivas (@ramesss) has provided another chart on Championship full backs, this time showing Yiadom as significantly different to Blackett/Gunter, with the second highest successful dribbles of all full/wing backs in the league, at 4.1 a game. Gunter and Blackett both hover just below the league average.



This combined with his lower-than-league-average numbers for (i) number of crossed per game and (ii) passes in the final 3rd (charts in earlier posts) suggests he is a (very) effective ball carrier but less active in the attacking department. Still, the ability to move the ball out from the back makes him a very useful asset, and is more consistent with the narrative of him standing out vs the other full backs, contrary to the potential new signing Halo effect discussed above.

User avatar
tmesis
Hob Nob Regular
Posts: 2780
Joined: 16 Aug 2013 20:26

Re: The only slightly more advanced than shots/possession statistics thread

by tmesis » 27 Apr 2019 11:34

muirinho
Brain Traysers ....

A bit like a relationship, all lovey-dovey to begin with, but after a couple of years, it's the teabags left in the sink, and clothes not put away that come to the fore.


Why does anyone leave teabags in the sink?


muirinho
Hob Nob Regular
Posts: 2075
Joined: 20 Jan 2016 12:10

Re: The only slightly more advanced than shots/possession statistics thread

by muirinho » 27 Apr 2019 11:35

tmesis
muirinho
Brain Traysers ....

A bit like a relationship, all lovey-dovey to begin with, but after a couple of years, it's the teabags left in the sink, and clothes not put away that come to the fore.


Why does anyone leave teabags in the sink?


I don't care why, I just want it to stop.

URZZZZ
Hob Nob Addict
Posts: 7279
Joined: 20 Apr 2013 18:30

Re: The only slightly more advanced than shots/possession statistics thread

by URZZZZ » 27 Apr 2019 12:25

Brain Traysers Ram Srinivas (@ramesss) has provided another chart on Championship full backs, this time showing Yiadom as significantly different to Blackett/Gunter, with the second highest successful dribbles of all full/wing backs in the league, at 4.1 a game. Gunter and Blackett both hover just below the league average.



This combined with his lower-than-league-average numbers for (i) number of crossed per game and (ii) passes in the final 3rd (charts in earlier posts) suggests he is a (very) effective ball carrier but less active in the attacking department. Still, the ability to move the ball out from the back makes him a very useful asset, and is more consistent with the narrative of him standing out vs the other full backs, contrary to the potential new signing Halo effect discussed above.


This is why I get slightly confused when we play 5 at the back, playing Gunter and Blackett as wing backs doesn't work. They both need a winger in front of them to act as the "link up" player because they don't have the ability to take the ball and dribble through players. Hence, Wigan away earlier this season when we played our narrow 4-3-2-1 formation with Gunter and Blackett as the fullbacks we had no threat from out wide. Even the PO final there was no real threat from them on either flank

Personally think it should be Yiadom and Blackett as the two wingbacks when we played that formation with Gunter slotting in as the third CB or Gunter as LWB and push Blackett as one of the three CB's. Personally think the two he plays at the moment is the worst combination but what do I know

As for your other point, I know he has quite a few assists but to me, he rarely swings in a decent cross, he always seems quite awkward when in the opposition box with a real lack of composure. But, as someone pointed out, he is top of our assists so I'm probably missing something

muirinho
Hob Nob Regular
Posts: 2075
Joined: 20 Jan 2016 12:10

Re: The only slightly more advanced than shots/possession statistics thread

by muirinho » 27 Apr 2019 13:02

URZZZZ
Brain Traysers Ram Srinivas (@ramesss) has provided another chart on Championship full backs, this time showing Yiadom as significantly different to Blackett/Gunter, with the second highest successful dribbles of all full/wing backs in the league, at 4.1 a game. Gunter and Blackett both hover just below the league average.



This combined with his lower-than-league-average numbers for (i) number of crossed per game and (ii) passes in the final 3rd (charts in earlier posts) suggests he is a (very) effective ball carrier but less active in the attacking department. Still, the ability to move the ball out from the back makes him a very useful asset, and is more consistent with the narrative of him standing out vs the other full backs, contrary to the potential new signing Halo effect discussed above.


This is why I get slightly confused when we play 5 at the back, playing Gunter and Blackett as wing backs doesn't work. They both need a winger in front of them to act as the "link up" player because they don't have the ability to take the ball and dribble through players. Hence, Wigan away earlier this season when we played our narrow 4-3-2-1 formation with Gunter and Blackett as the fullbacks we had no threat from out wide. Even the PO final there was no real threat from them on either flank

Personally think it should be Yiadom and Blackett as the two wingbacks when we played that formation with Gunter slotting in as the third CB or Gunter as LWB and push Blackett as one of the three CB's. Personally think the two he plays at the moment is the worst combination but what do I know

As for your other point, I know he has quite a few assists but to me, he rarely swings in a decent cross, he always seems quite awkward when in the opposition box with a real lack of composure. But, as someone pointed out, he is top of our assists so I'm probably missing something


He doesn't swing in crosses, he cuts inside and passes (or shoots & gets blocked). Whereas Gunter swings in crosses, and doesn't cut inside. A full back who does both competently is a rare thing in the Championship.

Brain Traysers
Member
Posts: 663
Joined: 06 Jul 2004 12:02
Location: Cascadia

Re: The only slightly more advanced than shots/possession statistics thread

by Brain Traysers » 23 May 2019 18:10

For people still uncertain about how the right application of analytics can benefit football, there is an excellent* long read on the NYT on how Liverpool's success in recent years has been supported by their research department. They have been the smartest club in the country, which is the only reason they have been anywhere near Man City.

*- as long as you don't mind being slightly patronised by some football basics - its clearly written for a US audience!


Brain Traysers
Member
Posts: 663
Joined: 06 Jul 2004 12:02
Location: Cascadia

Re: The only slightly more advanced than shots/possession statistics thread

by Brain Traysers » 23 May 2019 18:36

I also posted a number of scatter charts profiling full backs, which promoted a good debate over POTS candidate Yiadom. Similar charts have now been built into a nice article titled Profiling Ball-Playing Centre-backs in the Championship by Ram Srinivas and RFC fan PannasandNutmegs

Brain Traysers
Member
Posts: 663
Joined: 06 Jul 2004 12:02
Location: Cascadia

Re: The only slightly more advanced than shots/possession statistics thread

by Brain Traysers » 23 May 2019 18:42

And for my hat trick (I'm for for a bit now, I promise), here is more evidence of xG/analytics making its way more and more into mainstream media.

Statsbomb (one of the leading and probably the best football event data providers and consultants) have been looking back at key games from Man Utds 1999 Treble winning season through an xG lens.

Here is their assessment of the FA Cup semi final against Arsenal, featuring that Giggs goal/celebration (xG of 0.07).

And here is the analysis of the final against Newcastle.

Brain Traysers
Member
Posts: 663
Joined: 06 Jul 2004 12:02
Location: Cascadia

Re: The only slightly more advanced than shots/possession statistics thread

by Brain Traysers » 25 Jul 2019 22:44

A short little piece scratching the surface into the world of set piece analysis showing the scoring locations against the most popular defensive set up against corners:
https://twitter.com/From_The_Wing/status/1154438595354210306

(P.S. nice to see some analytics leaking into discussions on other threads :D )

Hound
Hob Nob Legend
Posts: 24934
Joined: 27 Sep 2016 22:16
Location: Simpleton

Re: The only slightly more advanced than shots/possession statistics thread

by Hound » 26 Jul 2019 10:05

thanks Brain. Not sure I'll get a chance to look through all, but I do find the stats and analysis interesting. Those two you mentioned on twitter, Panna' and Ram S are excellent at giving very quick visuals into the strengths of a player. Really enjoy looking at their work. Maybe you should all meet up for a geek session beer sometime

Brain Traysers
Member
Posts: 663
Joined: 06 Jul 2004 12:02
Location: Cascadia

Re: The only slightly more advanced than shots/possession statistics thread

by Brain Traysers » 29 Nov 2019 01:26

It's been a while since I last looked in detail at how our underlying numbers were looking this season, and conveniently Oli Walker has done a better job than I could in a StatsBomb article "Diving to the depths of the Championship relegation battle"

Firstly, its very refreshing to not be mentioned explicitly in an article about the Championship relegation battle.

According to the StatsBomb data (which is probably one of the better data sources - certainly their xG model is more advanced by capturing things like pressure by surrounding defenders) after 17 games this season (i.e. before Leeds) we ranked:
- 6th best for Non-Penalty xG difference
- 9th best for passes inside the box conceded
- 7th best for Non-Penalty xG Conceded
- 8th best for Open Play xG
- 10th best for xG/Shot - I find this most interesting as before Bowen we were shooting (and scoring) quite a few from outside the area

Certainly not relegation battle numbers, just things not clicking on the pitch so far

Hound
Hob Nob Legend
Posts: 24934
Joined: 27 Sep 2016 22:16
Location: Simpleton

Re: The only slightly more advanced than shots/possession statistics thread

by Hound » 29 Nov 2019 08:39

Brain Traysers It's been a while since I last looked in detail at how our underlying numbers were looking this season, and conveniently Oli Walker has done a better job than I could in a StatsBomb article "Diving to the depths of the Championship relegation battle"

Firstly, its very refreshing to not be mentioned explicitly in an article about the Championship relegation battle.

According to the StatsBomb data (which is probably one of the better data sources - certainly their xG model is more advanced by capturing things like pressure by surrounding defenders) after 17 games this season (i.e. before Leeds) we ranked:
- 6th best for Non-Penalty xG difference
- 9th best for passes inside the box conceded
- 7th best for Non-Penalty xG Conceded
- 8th best for Open Play xG
- 10th best for xG/Shot - I find this most interesting as before Bowen we were shooting (and scoring) quite a few from outside the area

Certainly not relegation battle numbers, just things not clicking on the pitch so far


how do we compare this season (and the previous one) compared to last in terms of the XG - assuming its a fair bit better

Brain Traysers
Member
Posts: 663
Joined: 06 Jul 2004 12:02
Location: Cascadia

Re: The only slightly more advanced than shots/possession statistics thread

by Brain Traysers » 29 Nov 2019 23:34

Hound
how do we compare this season (and the previous one) compared to last in terms of the XG - assuming its a fair bit better


The numbers vary by model, but the easiest dataset to answer that with is the fivethirtyeight, which is slightly more pessimistic this season than the figures from infogol, wyscout or statsbomb.

They have averaging -0.1 xG difference per game (1.1 for and 1.2 xG conceded each game), a significant improvement from -0.78 last season and -0.45 in 17/18. Our xG difference figures are 15th best in the league (sometimes referred to as the "justice league table"), a huge improvement on 23rd last season (only ahead of Ipswich), and 22nd in 17/18.

The data suggest we have improved at both ends of the pitch, with xG for up from 0.9 to 1.1 per game compared to last season, while xG against has dropped from 1.7 to 1.2


Who is online

Users browsing this forum: ankeny, Four Of Clubs, WestYorksRoyal and 420 guests

It is currently 28 Mar 2024 20:20