Top 10s - early season stats

[Davefevs]

[ExiledAjax]

@Davefevs interesting to see early on. I think it's also worth comparing this kind of end-product analysis to last season.

Shots are up from 7-8 per game to around 11. Still shy of the 15 you'd want to see if expecting promotion, but a 40-50% up tick isn't to be sniffed at. We've also managed to keep roughly the same % of our shots on target - we were at 33% under Pearson last term, and are at 31% in the four games so far. xG per shot is about the same as well, I see a small drop, but it's about what it was. So that means we've managed to increase the number of shots taken without sacrificing much in terms of quality. 

Thats reflected in the xG, where we definitely see a big improvement. We're at an ok average of around 1.3 per game (depending on source). Ideally you'd want 1.6ish, but compare it to last season, where at one point, we were as low as 0.75 (which is utterly shocking) and the improvement is huge.

xG against is also better, down from 1.82 under Pearson to 1.26 this season. Much better. I think the concession of relatively "soft" or "silly" goals supports that. 

Overall I have our average xG delta as only 0.04 per game across the four league games. Now that's an enormous improvement on the -1.05 we had under Pearson last season, and on the -0.41 under the whole of Holden's reign, but 0.04 doesn't support a prediction of winning many games. However, it should see us survive comfortably, and likely achieve the lower mid table prediction that seemed to be the general consensus pre-season.

I'd be interested to see how our xG against compares to the rest of the league.

[Bs4Red]

Personally think xG is one of the most ridiculous stats in football. 

Ive done lots of research and spoke to many coaches who agree however there are many who use it a lot too. 

 

[Red Skin]

I like xG as a measure of how attacking a team is.  I remember a couple of seasons ago where Brentford started the season with a small point haul, but on xG they were one of the top teams. 

Sure enough come the end of the season they had converted xG into actual goals and finished the season at the top end exactly where the xG suggested they would be.  

[Pezo]

4 hours ago, Bs4Red said:

Personally think xG is one of the most ridiculous stats in football. 

Ive done lots of research and spoke to many coaches who agree however there are many who use it a lot too. 

 

Would you prefer our xG matched or was worse than last seasons?

It's a weapon in the managers arsenal, I can see why a trainer/coach would dismiss it - it unlikely to help them organise a training plan.

It helps explain problems and there is a strong correlation between some analysis and finishing league positions, this helps in figuring out if you are doing better than you probably should be or worse than you probably should be. 

[Davefevs]

6 hours ago, ExiledAjax said:

@Davefevs interesting to see early on. I think it's also worth comparing this kind of end-product analysis to last season.

Shots are up from 7-8 per game to around 11. Still shy of the 15 you'd want to see if expecting promotion, but a 40-50% up tick isn't to be sniffed at. We've also managed to keep roughly the same % of our shots on target - we were at 33% under Pearson last term, and are at 31% in the four games so far. xG per shot is about the same as well, I see a small drop, but it's about what it was. So that means we've managed to increase the number of shots taken without sacrificing much in terms of quality. 

Thats reflected in the xG, where we definitely see a big improvement. We're at an ok average of around 1.3 per game (depending on source). Ideally you'd want 1.6ish, but compare it to last season, where at one point, we were as low as 0.75 (which is utterly shocking) and the improvement is huge.

xG against is also better, down from 1.82 under Pearson to 1.26 this season. Much better. I think the concession of relatively "soft" or "silly" goals supports that. 

Overall I have our average xG delta as only 0.04 per game across the four league games. Now that's an enormous improvement on the -1.05 we had under Pearson last season, and on the -0.41 under the whole of Holden's reign, but 0.04 doesn't support a prediction of winning many games. However, it should see us survive comfortably, and likely achieve the lower mid table prediction that seemed to be the general consensus pre-season.

I'd be interested to see how our xG against compares to the rest of the league.

XG and XGA for you.

[Davefevs]

5 hours ago, Bs4Red said:

Personally think xG is one of the most ridiculous stats in football. 

Ive done lots of research and spoke to many coaches who agree however there are many who use it a lot too. 

 

It’s far from perfect, but it’s tonnes better than Shots, or Shots on target taken in isolation….which seems to be still (badly imho) used as a qualifier for who should’ve won a game.

Is a 10 yard volley from the middle of the goal that smacks the cross bar worse than a 35 yard dribbler than the keeper can throw his cap on?  That’s where xG is trying to show “quality of chance” not number of chances.

[Fordy62]

2 minutes ago, Davefevs said:

It’s far from perfect, but it’s tonnes better than Shots, or Shots on target taken in isolation….which seems to be still (badly imho) used as a qualifier for who should’ve won a game.

Is a 10 yard volley from the middle of the goal that smacks the cross bar worse than a 35 yard dribbler than the keeper can throw his cap on?  That’s where xG is trying to show “quality of chance” not number of chances.

Thanks Dave. Stats that appear to match my feelings that we’re on the up. 

[petehinton]

Anything around successful tackles? Saw something the other day saying Palmer was our best tackler so far this season….

[Bristol is red]

Much better so far this season, something we need to keep up and improve on as the season goes on.  We had a horrible xG last season and it showed.    I'd be interested in defensive stats as that seems our achilles heel right now with those lapses of concentration.

[Davefevs]

17 minutes ago, petehinton said:

Anything around successful tackles? Saw something the other day saying Palmer was our best tackler so far this season….

Wyscout does sliding tackles so not a decent comparison, but here’s some promising ones.

 

[Bs4Red]

26 minutes ago, Davefevs said:

It’s far from perfect, but it’s tonnes better than Shots, or Shots on target taken in isolation….which seems to be still (badly imho) used as a qualifier for who should’ve won a game.

Is a 10 yard volley from the middle of the goal that smacks the cross bar worse than a 35 yard dribbler than the keeper can throw his cap on?  That’s where xG is trying to show “quality of chance” not number of chances.

I understand what it’s trying to show, I’ve spent a lot of time using it and understanding it. 

I still believe it’s very opinion based because what some feel is a big chance isn’t the same for others and so many variables come into play. 

I agree on shots on target. Typical “we had 9 shots on target and didn’t score” all from 25 yards or half chances. 

 

Watching the other night for me bar Martins chance I would say none of the others were big chances or ones we should score. 

Thats where xG needs a lot of improvement 

[Bs4Red]

1 minute ago, Davefevs said:

Wyscout does sliding tackles so not a decent comparison, but here’s some promising ones.

 

Again very subjective because it would show we do a lot of defending if you look at it from another point of view 

[Bs4Red]

1 hour ago, Pezo said:

Would you prefer our xG matched or was worse than last seasons?

It's a weapon in the managers arsenal, I can see why a trainer/coach would dismiss it - it unlikely to help them organise a training plan.

It helps explain problems and there is a strong correlation between some analysis and finishing league positions, this helps in figuring out if you are doing better than you probably should be or worse than you probably should be. 

I think you can watch a game and see whether you are better or worse off. Nothing in football is best for analysing than watching. 

Again i understand how it works, I just dont agree with its effectiveness as it is subjective and too many variants 

[Davefevs]

Just now, Bs4Red said:

I understand what it’s trying to show, I’ve spent a lot of time using it and understanding it. 

I still believe it’s very opinion based because what some feel is a big chance isn’t the same for others and so many variables come into play. 

I agree on shots on target. Typical “we had 9 shots on target and didn’t score” all from 25 yards or half chances. 

 

Watching the other night for me bar Martins chance I would say none of the others were big chances or ones we should score. 

Thats where xG needs a lot of improvement 

A big chance on the old Statszone app (I miss it because it’s real time) was xG over 0.30.

Matty James (the one in my video clip) was 0.24

Martin’s was 0.48

Weimann’s (the one where he flicks at James’s shot) was 0.43

Pirot’s goal was 0.23 (0.07 for the shot that hit the post)

Pirot’s shot saved by Bents was 0.26

[Davefevs]

3 minutes ago, Bs4Red said:

Again very subjective because it would show we do a lot of defending if you look at it from another point of view 

Yes, which in fairness this season we haven’t.  Much depends on the context, e.g. where does it happen?

If I resort on number of duels (not success %), we get a different picture.  None of the individuals I highlighted above appear in the top 30 based on volume….which is good.

1 minute ago, Bs4Red said:

I think you can watch a game and see whether you are better or worse off. Nothing in football is best for analysing than watching. 

Again i understand how it works, I just dont agree with its effectiveness as it is subjective and too many variants 

totally agree.  I take the Thomas Frank view…..use your eyes and then check the data marries up.

It doesn’t always, we aren’t professional analysts, we all have unconscious bias too.  I do spend a fair bit of the game checking shape, distances back to front, watching movement, not always the ball.

What is so much better imho is that James and King / Massengo don’t just sit in front of our CBs.  They are much more in touch with our attackers and therefore our attacks.

[Bs4Red]

7 minutes ago, Davefevs said:

A big chance on the old Statszone app (I miss it because it’s real time) was xG over 0.30.

Matty James (the one in my video clip) was 0.24

Martin’s was 0.48

Weimann’s (the one where he flicks at James’s shot) was 0.43

Pirot’s goal was 0.23 (0.07 for the shot that hit the post)

Pirot’s shot saved by Bents was 0.26

Yeah that’s what I mean Pirots goal (second shot) was literally an open net from 18 yards. 

Weimann flick was a far far less likely chance than Martins 

As I said it’s very subjective and that’s why I feel it isn’t great to use. 

I do think as I said before what you see with your eyes is far more intuitive than stats you can read. 

[Bs4Red]

3 minutes ago, Davefevs said:

Yes, which in fairness this season we haven’t.  Much depends on the context, e.g. where does it happen?

If I resort on number of duels (not success %), we get a different picture.  None of the individuals I highlighted above appear in the top 30 based on volume….which is good.

totally agree.  I take the Thomas Frank view…..use your eyes and then check the data marries up.

It doesn’t always, we aren’t professional analysts, we all have unconscious bias too.  I do spend a fair bit of the game checking shape, distances back to front, watching movement, not always the ball.

What is so much better imho is that James and King / Massengo don’t just sit in front of our CBs.  They are much more in touch with our attackers and therefore our attacks.

Exactly the same as me, I always watch the game with a coaches view and enjoy the little things. 

Yeah agree James seems a far more B2B than CDM. 

[Pezo]

22 minutes ago, Bs4Red said:

I think you can watch a game and see whether you are better or worse off. Nothing in football is best for analysing than watching. 

Again i understand how it works, I just dont agree with its effectiveness as it is subjective and too many variants 

But you can't watch all games that are going on at the same time whereas you can look at the statistics to figure out if you need to a closer look.

Think of it as a timesaving mechanism. 

[IAmNick]

1 hour ago, Bs4Red said:

As I said it’s very subjective and that’s why I feel it isn’t great to use. 

I do think as I said before what you see with your eyes is far more intuitive than stats you can read.

I'm interested - with xG which bit are you saying is subjective?

Personally, I think there's a bit of a contradiction here if we have an issue with things being subjective.

I'd say your eyes and own interpretation is more subjective than a statistics based model such as xG. Your eyes may be more accurate (you also have the knowledge of the outcome to factor in of course, so we can never be unbiased as a model can), but we are still more subjective by nature surely?

[ExiledAjax]

3 hours ago, Davefevs said:

XG and XGA for you.

Right, so 5.94 for, versus 5.28 against. 6th place for, solid mid-table against. A difference of 0.66, over 4 games, so +0.17 per game. As I say, it's much better than last season but not exactly world-beating.

Pretty close to what I have from my (now updated) average of Infogol, E361, and FootyStats' xG numbers. I have 5.60 total for, 4.8 total against, a difference of +0.20 per game.

9 hours ago, Bs4Red said:

Personally think xG is one of the most ridiculous stats in football. 

Ive done lots of research and spoke to many coaches who agree however there are many who use it a lot too. 

This is why I take xG from 3 sources and average them to give the numbers I then use to see how we are doing. It's not perfect, but for a guy looking at this as a hobby, I'm happy with it. I've used these figures to pretty accurately predict when we are and aren't going to get relegated/promoted the past few seasons. For example after those games last season when we were riding high? The numbers were never pointing at promotion. We had 17 points after 10 games, but average xG differential was only 0.1 - which is nowhere near what a top 6 team would have.

3 hours ago, Bs4Red said:

Watching the other night for me bar Martins chance I would say none of the others were big chances or ones we should score. 

Infogol pretty much agrees with this. Here's the shot map, size of the circle indicates percentage chance to score. Martin's early chance is the massive circle - and is rated as 54%. You can see the rest are rated very low, mostly below 15%. Weimann's 69th minute chance is at 23%.

 

[Son of Fred]

12 league games at home without a win....

Club record..

I hate stats.

[WarksRobin]

Time to get a petition together for the EFL to abandon points and base league tables on xG

[DaveInSA]

Also XG doesn't take into account that "Andi couldn't hit a barn door with a banjo" Weimann, couldn't hit a barn door with a banjo.

 

[Bs4Red]

15 hours ago, IAmNick said:

I'm interested - with xG which bit are you saying is subjective?

Personally, I think there's a bit of a contradiction here if we have an issue with things being subjective.

I'd say your eyes and own interpretation is more subjective than a statistics based model such as xG. Your eyes may be more accurate (you also have the knowledge of the outcome to factor in of course, so we can never be unbiased as a model can), but we are still more subjective by nature surely?

Yeah I would also agree with that. Our views are subjective but we also have a lot more information.

xG found that Pirot shot at goal was 0.28 or something. It can be found above however anyone who was watching would say it was an open net from 18 yards and you’d say he scores 90/100 times. 

Martins chance was less likely than Weimanns flick. Again anyone who watched would see it was a lot better chance. 

I guess for me as a coach I don’t find it useful and find my eyes better. But I also understand xG and what it’s trying to achieve. I just think it’s very flawed but again that’s my opinion. 

I respect people who use it I just don’t think it’s that useful a tool but I believe it could be. 

[Davefevs]

32 minutes ago, Bs4Red said:

Yeah I would also agree with that. Our views are subjective but we also have a lot more information.

xG found that Pirot shot at goal was 0.28 or something. It can be found above however anyone who was watching would say it was an open net from 18 yards and you’d say he scores 90/100 times. 

Martins chance was less likely than Weimanns flick. Again anyone who watched would see it was a lot better chance. 

I guess for me as a coach I don’t find it useful and find my eyes better. But I also understand xG and what it’s trying to achieve. I just think it’s very flawed but again that’s my opinion. 

I respect people who use it I just don’t think it’s that useful a tool but I believe it could be. 

Over a bigger sample, rather than taking individual chances it will average out well, because that is the underlying model.  The penalty kick is 0.76 xG… because 76 pens out of 100 are scored.

To get a big enough sample size you have to dumb down some of the variables, but most models take things like keeper position, under challenge, defenders between goal and ball, etc.  So although you might not see many Martin 4 yards out sliding chances you will see similar(ish) positions, etc.

 

[Shaun Taylor]

Anyone know how to find out how many away fans attended games in all four leagues.

[Major Isewater]

On 23/08/2021 at 09:19, Son of Fred said:

12 league games at home without a win....

Club record..

I hate stats.

[Mr Popodopolous]

On 22/08/2021 at 20:59, Bs4Red said:

I think you can watch a game and see whether you are better or worse off. Nothing in football is best for analysing than watching. 

Again i understand how it works, I just dont agree with its effectiveness as it is subjective and too many variants 

Brentford were always the side who outperformed their average but eventually came good.

They were dominant in many aspects over recent seasons, but this was not necessarily reflected with results- think Leeds under Bielsa were also dominant and though they missed it in year one, they eventually comfortably in the end- albeit was a battle until the final month/6 weeks, won this League. 

Whether they are perfect or not, they can be an indicator of things to come- when we were outperforming our xG and other underlying stats, slowly but surely our results started to catch up or stats with results depending on POV and the trajectory was downward. Our results at times felt unsustainable in 2019/20 and post the first 10 maybe, last season too- many other factors of course but like I say can be a useful indicator of what is to come.

[JamesBCFC]

On 23/08/2021 at 12:28, DaveInSA said:

Also XG doesn't take into account that "Andi couldn't hit a barn door with a banjo" Weimann, couldn't hit a barn door with a banjo.

 

Fortunately he's not paid to hit barn doors with banjos.