United and xG (now that Ole is gone will things change?)

Borys

Statistics Wizard
Joined
May 10, 2013
Messages
9,023
Location
Bielsko Biala, Poland
That shouldn't be included really then.
Why not?

Cummulative xG is not really any indicator. If you want to see if team overachieves you need to compare xG-Goals for each game, which is easy to do. Just because we have +7 goals on that parameter doesn't mean we score one extra goal (than expected) every 3rd game.
 

bosnian_red

Worst scout to ever exist
Joined
Aug 13, 2011
Messages
57,577
Location
Canada
Why not?

Cummulative xG is not really any indicator. If you want to see if team overachieves you need to compare xG-Goals for each game, which is easy to do. Just because we have +7 goals on that parameter doesn't mean we score one extra goal (than expected) every 3rd game.
People get weirdly offended by stats. I don't get it. It's literally just breaking down reality into values to help analyze. There's nothing really to dispute, you can argue interpretations of it but the statistic is the statistic.
 

bosnian_red

Worst scout to ever exist
Joined
Aug 13, 2011
Messages
57,577
Location
Canada
XG is an utterly bullshit “statistic”.
It's more accurate as a long term representation of a teams chance creation or attacking quality (and better used to extrapolate future regression) than total shots/shots on target or even the actual goals over a certain period of time is... so it really isn't.
 

Rado_N

Yaaas Broncos!
Joined
Apr 6, 2009
Messages
110,905
Location
Manchester
It's more accurate as a long term representation of a teams chance creation or attacking quality (and better used to extrapolate future regression) than total shots/shots on target or even the actual goals over a certain period of time is... so it really isn't.
It’s hugely subjective and inconsistent.

The use of stats in an argument will always be subjective in terms of how they’re interpreted, but individual stats themselves should be measurable and objective data points.

Player X completed 13 forward passes, or Player Y scored 8 goals in 2020, those are stats.

Player Z was through on goal, one guy reckons he had a 27% chance of scoring and one guy reckons he had a 45% chance of scoring, that’s not a stat.
 

Borys

Statistics Wizard
Joined
May 10, 2013
Messages
9,023
Location
Bielsko Biala, Poland
People get weirdly offended by stats. I don't get it. It's literally just breaking down reality into values to help analyze. There's nothing really to dispute, you can argue interpretations of it but the statistic is the statistic.
Yeah, totally.

I think you've mistaken me with someone. Or was this reply to my post?

It’s hugely subjective and inconsistent.

The use of stats in an argument will always be subjective in terms of how they’re interpreted, but individual stats themselves should be measurable and objective data points.

Player X completed 13 forward passes, or Player Y scored 8 goals in 2020, those are stats.

Player Z was through on goal, one guy reckons he had a 27% chance of scoring and one guy reckons he had a 45% chance of scoring, that’s not a stat.
If you build a database of those parameters you can make really valid conclusions, because variation between 27-45% becomes insignificant on a larger sample.
 

mav_9me

Full Member
Joined
Feb 9, 2009
Messages
12,400
Why not? Everything is part of the data set and just shows if a team is clinical over a season or not. Relative to the chances we are making this season, we have been clinical.
In that particular tweet it skews the data as this was an outlier.
 

bosnian_red

Worst scout to ever exist
Joined
Aug 13, 2011
Messages
57,577
Location
Canada
In that particular tweet it skews the data as this was an outlier.
The tweet just showed the overall goals minus the xG on the season, you don't take out any outliers. Every team has multiple outliers every year. That's why you get over/under performance. Otherwise pretty much everyone would be right around xG, as it's pretty much the likelihood on average that a specific type of chance has been finished based on loads and loads of samples over the years. Scoring 9 from 5-6 xG is no different than scoring 2 from 1.2-1.3 xG.
 

bosnian_red

Worst scout to ever exist
Joined
Aug 13, 2011
Messages
57,577
Location
Canada
It’s hugely subjective and inconsistent.

The use of stats in an argument will always be subjective in terms of how they’re interpreted, but individual stats themselves should be measurable and objective data points.

Player X completed 13 forward passes, or Player Y scored 8 goals in 2020, those are stats.

Player Z was through on goal, one guy reckons he had a 27% chance of scoring and one guy reckons he had a 45% chance of scoring, that’s not a stat.
It's not a guy deciding a chance has a certain likelihood, it's a model that looks at historical data and trends of how often a specific type of chance has been scored, and its pretty much an average % likelihood. Every penalty is worth 0.76 xG, because roughly 3/4 of penalties are actually scored in real life.

It's a model that has proven to be statically relevant over a period of time, so it is an accurate representation of chances that actually turned into shots. What it fails to capture is all chances that don't lead to a shot, or you have outliers with big chances that would've been called back for offside but it didn't go in so nobody cared, but still showed up in xG as it's just taking everything overall. But that tends to be a small amount that isn't too significant to any totals over the years.
 

adexkola

Doesn't understand sportswashing.
Joined
Mar 17, 2008
Messages
48,027
Location
The CL is a glorified FA Cup set to music
Supports
orderly disembarking on planes
I'm not sure why people get so annoyed with the use of statistics in sports. There is too much information that is generated in a single game to process just with the eye test. Not to talk of multiple games over an entire season.

No statistic is perfect. But as in other fields, our ability to systematically extract more useful information from football matches is increasing all the time.

We literally have a thread where people admit their perceptions of players are skewed based on their bad performances coinciding with when these people tuned in... And yet you want me to rely on these people's impressions of players or tactics? And no, I can't watch 380 Premier league games every season... Thankfully stats exist to help form a(n imperfect yet continually improving) picture of what happened in a game.
 

rotherham_red

Full Member
Joined
Mar 12, 2005
Messages
7,407
Why not? Everything is part of the data set and just shows if a team is clinical over a season or not. Relative to the chances we are making this season, we have been clinical.
And yet, we have missed the most big chances in the league as well (though this might have changed now, it was mentioned after the Arsenal match).

This season has been weird af.
 

bosnian_red

Worst scout to ever exist
Joined
Aug 13, 2011
Messages
57,577
Location
Canada
And yet, we have missed the most big chances in the league as well (though this might have changed now, it was mentioned after the Arsenal match).

This season has been weird af.
Yup that's the annoying part. Rashford for example tends to miss a lot of easy chances but score a lot of hard chances and it tends to even out over time. Gives you a sense of bad finishing when it's just our goals are distributed weirdly. Another year we could be clinical in big chances but score very few "small chances". Tends to even out over time!
 

tomaldinho1

Full Member
Joined
Nov 26, 2015
Messages
17,367
I'm not sure why people get so annoyed with the use of statistics in sports. There is too much information that is generated in a single game to process just with the eye test. Not to talk of multiple games over an entire season.

No statistic is perfect. But as in other fields, our ability to systematically extract more useful information from football matches is increasing all the time.

We literally have a thread where people admit their perceptions of players are skewed based on their bad performances coinciding with when these people tuned in... And yet you want me to rely on these people's impressions of players or tactics? And no, I can't watch 380 Premier league games every season... Thankfully stats exist to help form a(n imperfect yet continually improving) picture of what happened in a game.
It gets heat on here because xG in particular highlights how bad we are when it comes to chance creation and therefore the thread usually descends into an ole out versus in conversation.

xG and xA are great to look at for a general trend, you can't rely on them for single instances but it's accurate when you look across a season for example. United at the moment really struggle to create good chances, it's something everyone can see and you constantly see it referenced in here and the xG doesn't just hint at that, it outright confirms it. I made this point the other day but we are 'best' at outperforming our xG in the whole league (United +9, Leicester & Palace + 6 and Soton + 5) and the 'best' of the top 13 clubs for xA.

This is why I don't think the answer is just to buy better players because you need the foundations first. We shouldn't hang our hat on xG, xA but it does hint at a deeper issue than just lacking the players.
 

anant

Correctly predicted Italy to win Euro 2020
Joined
Feb 28, 2015
Messages
8,258
It gets heat on here because xG in particular highlights how bad we are when it comes to chance creation and therefore the thread usually descends into an ole out versus in conversation.

xG and xA are great to look at for a general trend, you can't rely on them for single instances but it's accurate when you look across a season for example. United at the moment really struggle to create good chances, it's something everyone can see and you constantly see it referenced in here and the xG doesn't just hint at that, it outright confirms it. I made this point the other day but we are 'best' at outperforming our xG in the whole league (United +9, Leicester & Palace + 6 and Soton + 5) and the 'best' of the top 13 clubs for xA.

This is why I don't think the answer is just to buy better players because you need the foundations first. We shouldn't hang our hat on xG, xA but it does hint at a deeper issue than just lacking the players.
And this is where I disagree. xG is a great stat undoubtedly if used over the long term. I also agree that xG of any team will revert back to mean after a long enough period of time.

However, there are two ways of looking at any stat: 1st is relative to oneself. like you're doing. We're somewhat overachieving in goals scored and underachieving on the goals against column, and no one is going to bat an eyelid.

The 2nd one is however, what should be used to gauge our actual position - that is, what is our xG, xGA, xGD relative to other clubs in the league. And this will signify how strong you are relative to the league.In terms of xG, we're the 3rd best in the league, In terms of xGA, we're the 4th best in the league (as per understat).

Remove the 1st three games, and we'll be 2nd and 3rd best team in attack and defence respectively. Start from prior to Pool game where our form supposedly got worse, it's 2nd and 4th in attack and defence (although we're talking about just 8 games here).

The thing is overachieving the xG on it's own isn't a bad thing. The issue arises when it's put you in a false position in the league, like it did to us in the Mou season, or when Newcastle finished 5th in the league
 

lysglimt

Full Member
Scout
Joined
Jun 1, 2008
Messages
15,100
Well - nothing is worse than shots on target. One time has 5 shots from 40 yards which the goalkeeper saves without an effort - that's 5 shots on target. The other team hits the post 3 times, and the opponent make 5 last ditch tackles that otherwise would be a goal - no shots on target. Shots on target shows 5-0 :)
 

bosnian_red

Worst scout to ever exist
Joined
Aug 13, 2011
Messages
57,577
Location
Canada
It gets heat on here because xG in particular highlights how bad we are when it comes to chance creation and therefore the thread usually descends into an ole out versus in conversation.

xG and xA are great to look at for a general trend, you can't rely on them for single instances but it's accurate when you look across a season for example. United at the moment really struggle to create good chances, it's something everyone can see and you constantly see it referenced in here and the xG doesn't just hint at that, it outright confirms it. I made this point the other day but we are 'best' at outperforming our xG in the whole league (United +9, Leicester & Palace + 6 and Soton + 5) and the 'best' of the top 13 clubs for xA.

This is why I don't think the answer is just to buy better players because you need the foundations first. We shouldn't hang our hat on xG, xA but it does hint at a deeper issue than just lacking the players.
It goes both ways though. Better players lead to better xG, but so do better systems (and better xG will in the end with a big enough sample always lead to improved results and move goals).
Our xG isn't that bad anyway, it's improving every year so are our xPts. We're 3rd to City and Liverpool in xG, and 4th in xGA (we actually are the worst in the top 13 clubs for goals against - xGA, you read it wrong, meaning we are due some luck in the goals against department). 4th in xPTS too. Us being top scorers is obviously skewed, but better players will help increase the xG and get our actual goal total to be a more sustainable number and not a temporary run of form that will fall. By better players, I mean any natural right winger will be massive, as would a natural striker. Cavani has had limited minutes but he's one of the best in the league in xG/90, so a right winger that can create and a natural striker would go a long way to sorting out our attacking consistency issues.
 

bosnian_red

Worst scout to ever exist
Joined
Aug 13, 2011
Messages
57,577
Location
Canada
And this is where I disagree. xG is a great stat undoubtedly if used over the long term. I also agree that xG of any team will revert back to mean after a long enough period of time.

However, there are two ways of looking at any stat: 1st is relative to oneself. like you're doing. We're somewhat overachieving in goals scored and underachieving on the goals against column, and no one is going to bat an eyelid.

The 2nd one is however, what should be used to gauge our actual position - that is, what is our xG, xGA, xGD relative to other clubs in the league. And this will signify how strong you are relative to the league.In terms of xG, we're the 3rd best in the league, In terms of xGA, we're the 4th best in the league (as per understat).

Remove the 1st three games, and we'll be 2nd and 3rd best team in attack and defence respectively. Start from prior to Pool game where our form supposedly got worse, it's 2nd and 4th in attack and defence (although we're talking about just 8 games here).

The thing is overachieving the xG on it's own isn't a bad thing. The issue arises when it's put you in a false position in the league, like it did to us in the Mou season, or when Newcastle finished 5th in the league
Yup, completely agree. You look at 16/17 and compare it to 17/18. Understat had City as by far the best team by xG that season, but were severely underperforming in terms of turning dominance into results, while Chelsea and Spurs overperformed like crazy. Chelsea the next season dipped to down to their rough xG numbers (75 xPts to 68 xPts, while real points were 93 to 70), Spurs went from 75 to 76 xPts, but real points down from 86 to 77.

Or United in 17/18 having 81 points from an xPts of 62. Go forward a year and we got 66 points from just under 62 xPts. There's way more correlation for season after season performance and overall quality if you look at xG. Sure, overperformance can lead to trophies too. Football is a low scoring sport and a season is only 38 games. Brighton might get relegated due to bad finishing this season, where otherwise they should be comfortable top half. It's a useful tool to help analyze a team, and you can track changes between managers and long term growth, or impact over time from certain players (look at United the year before Bruno joined to the year after he joined for example).

Everyone can see it with their eyes, a team "grinding their way" to a title and then all the "what happened" questions when they inevitably drop off the next year. XG just helps quantify all that. Many of us were unhappy on 17/18 and felt it was a step back from 16/17, but we finished 2nd with 81 points so people weren't complaining loads. Turned out to be valid complaints, and xG was great at highlighting that there were in fact big issues.
 

bosnian_red

Worst scout to ever exist
Joined
Aug 13, 2011
Messages
57,577
Location
Canada
@bosnian_red I admire your patience. Very good posts in this thread.
Ha, thanks. There's always the extreme opinions when it comes to analytics in all sports, but I just don't see the arguments against it. Its just a stat that helps quantify what we all see with our eyes, but if the numbers don't say what people want, they argue the stat is wrong. All stats have a lot of variance over periods of time, xG is pretty much proven to be the most accurate stat with the least amount of variance in it. Every football coach will set up their team to create high quality chances and restrict high quality chances. Nobody will create a system around creating low quality chances and giving away high quality chances "because they have a world class player so they don't need to do more". It's illogical. XG quantifies how effective a team is doing that, so I'll never understand the arguments. "Oh but the chance came in garbage time of the match when they stopped playing and it didn't matter". It still happened!
 

tomaldinho1

Full Member
Joined
Nov 26, 2015
Messages
17,367
And this is where I disagree. xG is a great stat undoubtedly if used over the long term. I also agree that xG of any team will revert back to mean after a long enough period of time.

However, there are two ways of looking at any stat: 1st is relative to oneself. like you're doing. We're somewhat overachieving in goals scored and underachieving on the goals against column, and no one is going to bat an eyelid.

The 2nd one is however, what should be used to gauge our actual position - that is, what is our xG, xGA, xGD relative to other clubs in the league. And this will signify how strong you are relative to the league.In terms of xG, we're the 3rd best in the league, In terms of xGA, we're the 4th best in the league (as per understat).

Remove the 1st three games, and we'll be 2nd and 3rd best team in attack and defence respectively. Start from prior to Pool game where our form supposedly got worse, it's 2nd and 4th in attack and defence (although we're talking about just 8 games here).

The thing is overachieving the xG on it's own isn't a bad thing. The issue arises when it's put you in a false position in the league, like it did to us in the Mou season, or when Newcastle finished 5th in the league
I understand the value of the second point, which in turn backs up the 1st coincidentally. For example - look at our xG, xA, xGD stats versus the other teams (your option 1) and we are over performing and then look at the xG league table (as you say in option 2) and we are worse off. It makes sense that we're only slightly worse off (on xG table) because we have been overperforming and, as said, under-lying stats aren't everything and there are no prizes for winning the xG league table but it's generally an accurate overall reflection of how teams are doing.

I don't get the relevance of adding/subtracting league games though, this season is this season.

It goes both ways though. Better players lead to better xG, but so do better systems (and better xG will in the end with a big enough sample always lead to improved results and move goals).
Our xG isn't that bad anyway, it's improving every year so are our xPts. We're 3rd to City and Liverpool in xG, and 4th in xGA (we actually are the worst in the top 13 clubs for goals against - xGA, you read it wrong, meaning we are due some luck in the goals against department). 4th in xPTS too. Us being top scorers is obviously skewed, but better players will help increase the xG and get our actual goal total to be a more sustainable number and not a temporary run of form that will fall. By better players, I mean any natural right winger will be massive, as would a natural striker. Cavani has had limited minutes but he's one of the best in the league in xG/90, so a right winger that can create and a natural striker would go a long way to sorting out our attacking consistency issues.
Yes I'm not saying buying better players won't help but that's what we've always done since SAF and it's never been the solution. I don't know how useful it is to compare xG to previous years, for example the best it has been for us is comfortably in Mou's 16/17 year where we averaged out at 1.98 versus 1.81 this season and we didn't even get top four then! This year should be viewed in isolation (no pun intended) given covid and the mental results we saw early on.

Sorry that's what I meant re 13th, we are the worst of all those clubs. The issue I see is our defence is worse than it's been in a while (despite high value signings) and we aren't really offsetting that enough with goals. What would be interesting is to see our goals variance because I can't help but think the Soton and Leeds game are masking the issue somewhat as well. For me it's a case of let's get a setup functioning with the players we have - who for all our moaning are actually really good across all positions - and then you buy players to tweak and get to the elite level.
 
Last edited:

roonster09

Hercule Poirot of the scouting world
Scout
Joined
May 10, 2009
Messages
36,514
It gets heat on here because xG in particular highlights how bad we are when it comes to chance creation and therefore the thread usually descends into an ole out versus in conversation.
That's not really true. For example, @Rado_N wanted ole out and still thinks xG is a pile of shit.

This is the problem in the forum, trying to categorise everything into group a vs b.
 

MadDogg

Full Member
Joined
Apr 24, 2002
Messages
15,809
Location
Manchester Utd never lose, just run out of time
And yet, we have missed the most big chances in the league as well (though this might have changed now, it was mentioned after the Arsenal match).

This season has been weird af.
Yep. Bruno and Rashford have scored quite a lot of goals from difficult positions. Meanwhile Martial and Rashford tend to miss quite a lot of good chances.
 

Paul_Scholes18

Full Member
Joined
Sep 13, 2014
Messages
13,891
I think xG is interesting, but I would not read too much into total season stats with it.
Actually watching games give a better indication of what is going on in the games.
You can more easily see what is dangerous and how much pressure teams put on each other.
Of course it is hard to remember games 10 weeks ago so looking at stats can be a good way to remind ourselfes about them.

I think you need to analyse each games to really see how a team is doing.
Just like looking at total goals scores might not tell the full story with many goals scored in specific games and few in others. It is the same with xG as well.

Football is about getting the 3 points and not about scoring the most goals. City have been better at doing that than us recently and thus increased the lead.
They have great defensive stats too and probably are more dominant than they even would need to be in terms of not conceding many goals or chances. Liverpool last year is a good example of being dominant enough to win games, but not to smash teams regulary though and that is an equally good way of getting a title win.
 

Teja

Full Member
Joined
Aug 17, 2014
Messages
5,701
On the topic of data vs human judgement I think it has been proven time and time again that data will, if not outright win, have its place in any analysis. Humans have emotions, conscious / unconscious biases and can outright intentionally lie because of agendas. Data is never complete / rich enough to tell the whole story. So some tension here is good, but dismissing data with things like "lies, damned lies and statistics" as if that proved a point IMO is pretty idiotic. If you disagree with the data being shown, pointing out specifically why that is so instead of resorting to banalities.

What has specifically changed *now* is that we have data that's so much richer (and more useful) because of how computer vision, machine learning etc. are being applied to football analytics. For technical / data-sciency type folks I'd highly recommend checking out the statsbomb channel.

https://www.youtube.com/channel/UCmZ2ArreL9muPvH49Gaw0Bw/videos
Press resistant midfielder analysis: https://statsbomb.com/2021/02/statsbomb-data-case-studies-actions-under-pressure/
Pressing styles as viewed through data:
Liverpool and Google Deep Mind (of AlphaGo fame) recently published a paper with some specifics


I think it's inevitable that the data revolution is coming to football. There are people and clubs that accept it that'll move ahead of the pack and others that don't but will eventually be forced to.
 

anant

Correctly predicted Italy to win Euro 2020
Joined
Feb 28, 2015
Messages
8,258
I understand the value of the second point, which in turn backs up the 1st coincidentally. For example - look at our xG, xA, xGD stats versus the other teams (your option 1) and we are over performing and then look at the xG league table (as you say in option 2) and we are worse off. It makes sense that we're only slightly worse off (on xG table) because we have been overperforming and, as said, under-lying stats aren't everything and there are no prizes for winning the xG league table but it's generally an accurate overall reflection of how teams are doing.

I don't get the relevance of adding/subtracting league games though, this season is this season.



Yes I'm not saying buying better players won't help but that's what we've always done since SAF and it's never been the solution. I don't know how useful it is to compare xG to previous years, for example the best it has been for us is comfortably in Mou's 16/17 year where we averaged out at 1.98 versus 1.81 this season and we didn't even get top four then! This year should be viewed in isolation (no pun intended) given covid and the mental results we saw early on.

Sorry that's what I meant re 13th, we are the worst of all those clubs. The issue I see is our defence is worse than it's been in a while (despite high value signings) and we aren't really offsetting that enough with goals. What would be interesting is to see our goals variance because I can't help but think the Soton and Leeds game are masking the issue somewhat as well. For me it's a case of let's get a setup functioning with the players we have - who for all our moaning are actually really good across all positions - and then you buy players to tweak and get to the elite level.
So, even as per the xG table, we're 3rd (and I'll be using xG, xGA, xGD - not completely convinced by xPts). And that's what has been promising under Ole. Under Moyes, LVG, Mou - in each of the seasons as per chances created vs conceded we were between 5th-7th. However, we were 4th last season, 3rd this season, and that is progress.

The reason for looking at different time periods is that the league wasn't really fair in the 1st 3-4 games as every team didn't have similar length of pre-season, and so the chances that the season would pan out like those 3 games is unlikely.


Coming to your defence related points, we're a shaky defence and all, which I agree with. But the issue is not that we concede a lot of quality chances - our xGA since the Spurs game has been less than a goal/game - it's actually the fact that the shots that we conceded are often high quality (nPxG/shot is 0.1 for us which is on the higher side when compared to other teams) and most of these shots that we've conceded are avoidable and mostly due to concentration lapses and lack of communication
 

tomaldinho1

Full Member
Joined
Nov 26, 2015
Messages
17,367
That's not really true. For example, @Rado_N wanted ole out and still thinks xG is a pile of shit.

This is the problem in the forum, trying to categorise everything into group a vs b.
I didn't actually realise until now - the thread title is actually about Ole and xG so I guess it's unavoidable here. I'm all for more threads where it doesn't just descend into the usual posters arguing over the manager but providing one example doesn't negate the fact that is sadly what happens to many threads on the caf.

So, even as per the xG table, we're 3rd (and I'll be using xG, xGA, xGD - not completely convinced by xPts). And that's what has been promising under Ole. Under Moyes, LVG, Mou - in each of the seasons as per chances created vs conceded we were between 5th-7th. However, we were 4th last season, 3rd this season, and that is progress.

The reason for looking at different time periods is that the league wasn't really fair in the 1st 3-4 games as every team didn't have similar length of pre-season, and so the chances that the season would pan out like those 3 games is unlikely.


Coming to your defence related points, we're a shaky defence and all, which I agree with. But the issue is not that we concede a lot of quality chances - our xGA since the Spurs game has been less than a goal/game - it's actually the fact that the shots that we conceded are often high quality (nPxG/shot is 0.1 for us which is on the higher side when compared to other teams) and most of these shots that we've conceded are avoidable and mostly due to concentration lapses and lack of communication
That's why I want us to just concentrate on this season and not even think about who we should sign. Then we assess if there is progress and, if there is, we put our trust in Ole for another season. As you say, if we stay where we are and retain similar stats, it'll be overall a decent season & xG is a nice compliment to look at alongside league position. If we finish 2nd and have the same xG,xA etc. stats, a fair assumption is the defence is individually not good enough (probably true as you say when you combine it with nPxG) but the attack is individually good enough but not working properly (again I'd say fair because last season we saw Mason, Martial and Rashford all score a lot of goals).

The league is never fair though, City had their extra time off given the covid outbreak and have been great since, Liverpool had loads of injuries. My point is just the season is the season and we can do the whole 'league table since March' type stuff but, especially this season, I think we just take it at face value and don't try to chop/change the sample size.
 

Withnail

Full Member
Joined
Jan 5, 2019
Messages
29,327
Location
The Arena of the Unwell
Yup that's the annoying part. Rashford for example tends to miss a lot of easy chances but score a lot of hard chances and it tends to even out over time. Gives you a sense of bad finishing when it's just our goals are distributed weirdly. Another year we could be clinical in big chances but score very few "small chances". Tends to even out over time!
You say that but would you say the same about Salah?

They've both missed 9 big chances in the PL.
 

Revan

Assumptionman
Joined
Dec 19, 2011
Messages
49,419
Location
London
On the topic of data vs human judgement I think it has been proven time and time again that data will, if not outright win, have its place in any analysis. Humans have emotions, conscious / unconscious biases and can outright intentionally lie because of agendas. Data is never complete / rich enough to tell the whole story. So some tension here is good, but dismissing data with things like "lies, damned lies and statistics" as if that proved a point IMO is pretty idiotic. If you disagree with the data being shown, pointing out specifically why that is so instead of resorting to banalities.

What has specifically changed *now* is that we have data that's so much richer (and more useful) because of how computer vision, machine learning etc. are being applied to football analytics. For technical / data-sciency type folks I'd highly recommend checking out the statsbomb channel.

https://www.youtube.com/channel/UCmZ2ArreL9muPvH49Gaw0Bw/videos
Press resistant midfielder analysis: https://statsbomb.com/2021/02/statsbomb-data-case-studies-actions-under-pressure/
Pressing styles as viewed through data:
Liverpool and Google Deep Mind (of AlphaGo fame) recently published a paper with some specifics


I think it's inevitable that the data revolution is coming to football. There are people and clubs that accept it that'll move ahead of the pack and others that don't but will eventually be forced to.
Saved the paper to read later. DeepMind are very good (though also overhyped) and have done some great work (including the protein prediction a few months ago).
 

Withnail

Full Member
Joined
Jan 5, 2019
Messages
29,327
Location
The Arena of the Unwell
On the topic of data vs human judgement I think it has been proven time and time again that data will, if not outright win, have its place in any analysis. Humans have emotions, conscious / unconscious biases and can outright intentionally lie because of agendas. Data is never complete / rich enough to tell the whole story. So some tension here is good, but dismissing data with things like "lies, damned lies and statistics" as if that proved a point IMO is pretty idiotic. If you disagree with the data being shown, pointing out specifically why that is so instead of resorting to banalities.

What has specifically changed *now* is that we have data that's so much richer (and more useful) because of how computer vision, machine learning etc. are being applied to football analytics. For technical / data-sciency type folks I'd highly recommend checking out the statsbomb channel.

https://www.youtube.com/channel/UCmZ2ArreL9muPvH49Gaw0Bw/videos
Press resistant midfielder analysis: https://statsbomb.com/2021/02/statsbomb-data-case-studies-actions-under-pressure/
Pressing styles as viewed through data:
Liverpool and Google Deep Mind (of AlphaGo fame) recently published a paper with some specifics


I think it's inevitable that the data revolution is coming to football. There are people and clubs that accept it that'll move ahead of the pack and others that don't but will eventually be forced to.
Thanks, I'll have to give that a look later. I'd always go for the data although it's obviously open to interpretation.

Many people don't realise you can't actually trust your memory as much as you think you can. The 'eye test' can never trump data in my view. Eye witnesses are notoriously unreliable.
 

Crashoutcassius

Full Member
Joined
Oct 10, 2013
Messages
10,295
Location
playa del carmen
It’s hugely subjective and inconsistent.

The use of stats in an argument will always be subjective in terms of how they’re interpreted, but individual stats themselves should be measurable and objective data points.

Player X completed 13 forward passes, or Player Y scored 8 goals in 2020, those are stats.

Player Z was through on goal, one guy reckons he had a 27% chance of scoring and one guy reckons he had a 45% chance of scoring, that’s not a stat.
What if, over a long period of time, guy As model comes out at approx the % conversion that his model predicted
 

bosnian_red

Worst scout to ever exist
Joined
Aug 13, 2011
Messages
57,577
Location
Canada
You say that but would you say the same about Salah?

They've both missed 9 big chances in the PL.
You have to look at it as a percentage. Over Rashfords career in the league, his goals are at 99% of his xG, so marginally under. Salah is 116% of his xG, consistently scoring a few more. He might miss chances, but he also finishes far more and generally gets plenty.
 

TMDaines

Fun sponge.
Joined
Sep 1, 2014
Messages
13,906

Understat have us fortunate to get a point. We played well for long periods, all whilst being utterly toothless.
 

Withnail

Full Member
Joined
Jan 5, 2019
Messages
29,327
Location
The Arena of the Unwell

Understat have us fortunate to get a point. We played well for long periods, all whilst being utterly toothless.
Well yeah our final ball was poor and often the ball before what would have been the final ball went awry as well.

I'm not sure about fortunate. I think it was a fair result as Chelsea didn't really create enough to win it either.
 

TMDaines

Fun sponge.
Joined
Sep 1, 2014
Messages
13,906
Well yeah our final ball was poor and often the ball before what would have been the final ball went awry as well.

I'm not sure about fortunate. I think it was a fair result as Chelsea didn't really create enough to win it either.
That double chance was pretty big for Chelsea. They score there more often than not.

It's why I cannot get overly vexed at the injustice of not getting a penalty. We should have had one, but it was an unforced error and would have bailed us out of actually having to create any chances. We're too reliant on those sorts of things.