Ole and xG

Amarsdd

Full Member
Joined
Jun 7, 2013
Messages
558
I don't understand why people are questioning xG statistics itself. It is just estimating probabilities of chances based on 1000+ (if I'm correct) similar situations that have occurred before. It's a completely objective view of a team's performance in and around the goal, but maybe not so much about the domination of the game in and out of possession. Obviously, subjectively the performances might look different to different people.
 

0le

Full Member
Joined
Dec 25, 2017
Messages
2,185
I don't understand why people are questioning xG statistics itself. It is just estimating probabilities of chances based on 1000+ (if I'm correct) similar situations that have occurred before. It's a completely objective view of a team's performance in and around the goal, but maybe not so much about the domination of the game in and out of possession. Obviously, subjectively the performances might look different to different people.
Well said.
 

El Zoido

Full Member
Joined
Jun 7, 2013
Messages
7,620
Location
UK
I don't understand why people are questioning xG statistics itself. It is just estimating probabilities of chances based on 1000+ (if I'm correct) similar situations that have occurred before. It's a completely objective view of a team's performance in and around the goal, but maybe not so much about the domination of the game in and out of possession. Obviously, subjectively the performances might look different to different people.
People are questioning it because they hate Ole and want him to fail and want him sacked.
 

Eddy_JukeZ

Full Member
Joined
Aug 21, 2012
Messages
10,463
We’re playing better now than we were when we were on the unbeaten run when he first joined.
Our performance last season vs Cardiff was one of the best by a team all season.

Our PPDA that game was the highest from any team in Europe.

There's no way we're playing better now than the opening 2-3 games of that run(later on - I'd agree).
 

Eddy_JukeZ

Full Member
Joined
Aug 21, 2012
Messages
10,463
It is interesting that xG has us 2nd and basically underachieving so far. Though 2 pen misses likely attribute to that heavily.

We'll see how it looks after a larger sample size.

I don't think we've played amazing, but we've definitely throw away points.
 

cptkeane1993

New Member
Newbie
Joined
Aug 31, 2019
Messages
11
I’m looking forward to the regression to the mean... Ole is doing something right.
 

bosnian_red

Worst scout to ever exist
Joined
Aug 13, 2011
Messages
37,221
Location
Canada
We were extremely lucky in that series of 10+ games we won. Everything was screaming that it was a purple patch and the xG was supporting that. I said back then, that soon we will get back to reality, which then happened.

It is the opposite this time around. We are outperforming xG and soon the results will follow. The regress to the mean is inevitable, though to be fair you don't need xG to realise that. We have played far better than to get 5 points out of 12.

About xG itself it has been explained many times. It is simply a start which is assigned based on the results of shots from the same positions. Most xG stats are based on 100k+ shots, where each position a shot has been taken is given a probability of being a goal based on similar shots. Far from perfect and very primitive compared to learning techniques which are in other fields, but it is still by far the best predictor in football.
Even that's not really true. We played well - even with xG. It's just we were in a purple patch by showing title winning form, while then we hit a rough patch of underperforming xG. In reality, we were basically 3rd in xG for Ole's time, which was how it showed overall during his time. It was just split by 2 extremes instead of an average throughout the season.
 

romufc

Full Member
Joined
Apr 30, 2019
Messages
3,161
It doesnt matter, we could have the best xG but that doesn't correspond to points.

Goals and clean sheet does.
 

bosnian_red

Worst scout to ever exist
Joined
Aug 13, 2011
Messages
37,221
Location
Canada
It doesnt matter, we could have the best xG but that doesn't correspond to points.

Goals and clean sheet does.
All xG does is point towards where your performance levels will average out over a long period of time. You will to through purple patches of finishing and bad runs where the opposition is clinical while you arent, but play well often enough and itll show over time and you'll regress to the mean of your performance the vast majority of time. Rarely, drastic over/underperformance lasts a whole season (United in 17/18), but then the following season we went back to shit like our xG was showing with Mourinho. It's as good of an indicator for future long term performances as anything.
 

varga92

New Member
Newbie
Joined
Oct 11, 2017
Messages
111
First of all, 4 games is very small sample size for xG.

Secondly, we need to look these stats in context. 1 penality equals 0.76 xG. Our xG from open play is only 4.52 and we scored 6 goals from open play.

Our xGA from corner is 0.37 and they scored against us 2 goals from corner. That's a long-term problem. We cannot defend well these situations. And if I'm not wrong, Ole had similar problems at Cardiff.
 

ash_86

Full Member
Joined
Sep 4, 2013
Messages
5,614

We have the best expected goals conceded metrics in the league. So we should be conceding less and scoring more according to Xg but opposite is happening. Surely the tide should turn eventually in our favor ?
 

bucky

Full Member
Joined
Apr 5, 2011
Messages
8,287

We have the best expected goals conceded metrics in the league. So we should be conceding less and scoring more according to Xg but opposite is happening. Surely the tide should turn eventually in our favor ?
4 games is a small sample size though.
 

shabadu84

Mint? Berry?
Joined
Jan 12, 2010
Messages
4,606
Location
Muppet Treasure Island
We were able to outperform our xG in the past based on good finishing. Any regression to the mean is going to require some additional luck but, more importantly, significant improvement in finishing. Martial, Rashford, Lingard, etc all need to do more to put the ball in the net or we're going to need to find players that won't be as wasteful.
 

Treble

Full Member
Joined
Sep 11, 2008
Messages
10,643

We have the best expected goals conceded metrics in the league. So we should be conceding less and scoring more according to Xg but opposite is happening. Surely the tide should turn eventually in our favor ?
Confirms the impression that Liverpool have been a bit average at the back. It's a ridiculously small sample though. Note that the bottom 3 teams have played City. The latter create many chances and change your xGA, especially when the sample is small. We have played teams with average attacks.
 

ash_86

Full Member
Joined
Sep 4, 2013
Messages
5,614
Confirms the impression that Liverpool have been a bit average at the back. It's a ridiculously small sample though. Note that the bottom 3 teams have played City. The latter create many chances and change your xGA, especially when the sample is small. We have played teams with average attacks.
Yes, i see the argument.
 

lysglimt

Full Member
Joined
Jun 1, 2008
Messages
9,519
What have West Ham been doing ? Is it all because of the City-game or have they really been that poor defensively ?
 

Grande

Full Member
Joined
Jun 22, 2007
Messages
2,615
Location
The Land of Do-What-You-Will
I’m not too crazy about xG stats normally, like attempts on goal and possesion it is sometimes interesting if you already know the context of the game, or sometimes when you see a very surprising result in a large sample. But in the end it’s goals that count, doesn’t matter if you get them by being effective or by producing a lot of half-situations.

However it is more interesting in the context of a build up over time. This season, the points total to me will probably be less important to me than seeing if our general play is developping in a positive way. Then these stats are more interesting, and xG and xPts more so.

My general impression from watching the games, is that we’ve played not fantastically but fairly well in each game. Chelsea result lied a bit and Palace result lied a bit, and we could have gotten more from the two other games too with a bit of margines. I’d say we played slightly better than in many of the games of our golden run last season, and much better than in the run up.

XG seems to confirm this. Under Mou in the second place season, our ratio of xG for to xG against was about 3 to 2. In his half of last season it was almost 1 to 1. Which fits with my impression. Under Ole last season, in the golden phase it was first 3 to 1 in the first games, trailing off to about 3 to 2, while in the run up it was a little bit better than 1 to 1.

Now it is about 2 to 1, more or less the same as Liverpool, only City are better, and the rest of the big six are well behind. All this is without pens.

Albeit being a small sample, to me, in context, this confirms the impression from watching the games, that our general play looks to have improved and is slightly better than both Mourinho’s best season and most of Solskjær’s good run last season. And this is with the removal of several of the more experienced players (Herrera, Matic, Lukaku, Sanchez), have the youngest starting squad in the PL, and are bedding in a new back four.

It seems an indication that we can forget the disastrous end to last season for now, and that we may well be in the process of a slow but largely consistent improval.

Thoughts?
 

Gator Nate

Full Member
Joined
Jan 15, 2019
Messages
624
Location
Orlando, FL
Supports
Unaffiliated
I’m not too crazy about xG stats normally, like attempts on goal and possesion it is sometimes interesting if you already know the context of the game, or sometimes when you see a very surprising result in a large sample. But in the end it’s goals that count, doesn’t matter if you get them by being effective or by producing a lot of half-situations.

However it is more interesting in the context of a build up over time. This season, the points total to me will probably be less important to me than seeing if our general play is developping in a positive way. Then these stats are more interesting, and xG and xPts more so.

My general impression from watching the games, is that we’ve played not fantastically but fairly well in each game. Chelsea result lied a bit and Palace result lied a bit, and we could have gotten more from the two other games too with a bit of margines. I’d say we played slightly better than in many of the games of our golden run last season, and much better than in the run up.

XG seems to confirm this. Under Mou in the second place season, our ratio of xG for to xG against was about 3 to 2. In his half of last season it was almost 1 to 1. Which fits with my impression. Under Ole last season, in the golden phase it was first 3 to 1 in the first games, trailing off to about 3 to 2, while in the run up it was a little bit better than 1 to 1.

Now it is about 2 to 1, more or less the same as Liverpool, only City are better, and the rest of the big six are well behind. All this is without pens.

Albeit being a small sample, to me, in context, this confirms the impression from watching the games, that our general play looks to have improved and is slightly better than both Mourinho’s best season and most of Solskjær’s good run last season. And this is with the removal of several of the more experienced players (Herrera, Matic, Lukaku, Sanchez), have the youngest starting squad in the PL, and are bedding in a new back four.

It seems an indication that we can forget the disastrous end to last season for now, and that we may well be in the process of a slow but largely consistent improval.

Thoughts?
My first thought is... You're gonna get crucified for that. :lol:

As for my own untimely demise, that's an interesting take and I'd tend to agree. It's just unpopular to admit we're probably better than we were last season when you look at W-D-L-D in the first four games. While the Chelsea game probably pads the stats a bit, there are only two teams with a better goal differential right now, City (+11) and Pool (+9) as would be expected, and Leicester is tied with us at +3. Palace and Spurs are the only other teams with a positive GD right now at +1. Also, four games in and only Watford and Wolves haven't won a match yet.
 

Cassidy

No longer at risk of being mistaken for a Scouser
Joined
Oct 2, 2013
Messages
19,055
We were able to outperform our xG in the past based on good finishing. Any regression to the mean is going to require some additional luck but, more importantly, significant improvement in finishing. Martial, Rashford, Lingard, etc all need to do more to put the ball in the net or we're going to need to find players that won't be as wasteful.
Martial has hardly been wasteful. Lingard.. well. Will be interesting if James can be a real goal threat from the left because if he can that would be a great suprise
 

elmo

Can never have too many Eevees
Joined
Jul 14, 2008
Messages
7,365
Location
AKA: Slapanut Goat Smuggla
All xG does is point towards where your performance levels will average out over a long period of time. You will to through purple patches of finishing and bad runs where the opposition is clinical while you arent, but play well often enough and itll show over time and you'll regress to the mean of your performance the vast majority of time. Rarely, drastic over/underperformance lasts a whole season (United in 17/18), but then the following season we went back to shit like our xG was showing with Mourinho. It's as good of an indicator for future long term performances as anything.
xG is a terrible stat for predicting future long term performance unless you believe that teams stay the same over the years despite multiple transfers and promotion/relegations.

xG is a good idea in theory, but it just doesn't work in football when sample sizes are so low and teams don't play each other enough times for the numbers to be relevant at all.
 

elmo

Can never have too many Eevees
Joined
Jul 14, 2008
Messages
7,365
Location
AKA: Slapanut Goat Smuggla
It is interesting that xG has us 2nd and basically underachieving so far. Though 2 pen misses likely attribute to that heavily.

We'll see how it looks after a larger sample size.

I don't think we've played amazing, but we've definitely throw away points.
Which tells you how worthless this stat is because the actual play on the pitch has been subpar mostly.

Our position on the table reflects how we've played so far this season. Mediocre.
 

Cassidy

No longer at risk of being mistaken for a Scouser
Joined
Oct 2, 2013
Messages
19,055
Which tells you how worthless this stat is because the actual play on the pitch has been subpar mostly.

Our position on the table reflects how we've played so far this season. Mediocre.
It is not worthless. We should have scored the penalties and collected the points and we should have beaten Wolves regardless.

Our play deserved more points than we got even if we didnt play great.

It does show also that we are not really creating many chances or enough chances which is what we have seen
 

noodlehair

"It's like..."
Joined
Apr 1, 2004
Messages
12,490
Location
Flagg
I'm not sure what this sort of stuff proves. We've dropped points the last two games because we concede ridiculously cheap goals that allow teams to score against us without needing to create chances themselves or dominate parts of the game.

That isn't something that's going to fix itself just because we have good XG stats.

If you take the stupid errors and pathetic set piece play out of our games the performances have actually been fairly good, barring the first half of the Palace game, which was terrible.
 

billybee99

New Member
Newbie
Joined
Jul 27, 2015
Messages
511
Its not. Its a very good reflection of teams' perfomances over long periods.
Do you know what can also give you a good reflection of teams' performance over long periods? Your eyeballs; your brain. Try using them. Some of you xG guys need to close your laptops, climb the stairs from your mother's basement and go have a beer, meet some people, get laid, do something, anything.
 

Grande

Full Member
Joined
Jun 22, 2007
Messages
2,615
Location
The Land of Do-What-You-Will
I'm not sure what this sort of stuff proves. We've dropped points the last two games because we concede ridiculously cheap goals that allow teams to score against us without needing to create chances themselves or dominate parts of the game.

That isn't something that's going to fix itself just because we have good XG stats.

If you take the stupid errors and pathetic set piece play out of our games the performances have actually been fairly good, barring the first half of the Palace game, which was terrible.
It doesn’t prove anything, but it confirms exactly what you and I have seen: That our performances have generally been fairly good. Many people look at our points tally and goal results to judge the performances, and conclude that we’ve been bad, but in our present context, xG sames more relevant than these stats. Because the context now is that this season we are trying to build a team and establish a new playing style, not actually trying to win the league (in ehich case losing points when performances are good would be a more crucial flaw).

Gaffes by players in new constallations are easier to iron out than if we constantly let opponents through on david, which we’ve seen a bit of in previous seasons. I take it as a good sign, nothing more.
 

Gator Nate

Full Member
Joined
Jan 15, 2019
Messages
624
Location
Orlando, FL
Supports
Unaffiliated
Which tells you how worthless this stat is because the actual play on the pitch has been subpar mostly.

Our position on the table reflects how we've played so far this season. Mediocre.
See, that's where I disagree. Our finishing has been poor.

vs. Wolves - we win possession 66/34, we win shots 9/6, we tie shots on target 2/2 - game ends in draw

vs. Palace - we win possession 71/29, we win shots 22/5, we tie shots on target 3/3 - we lose 2-1

vs. Southhampton - we win possession 59/41, we win shots 21/10, we win shots on target 8/2 - game ends in draw

In three matches, we've done no worse than 59% possession, taken 52 shots with 21 shots on target and have 4 goals to show for it.

That's a lack of finishing. For all the criticism of what's going on on the field, 4 of 21 is the problem.
 

noodlehair

"It's like..."
Joined
Apr 1, 2004
Messages
12,490
Location
Flagg
It doesn’t prove anything, but it confirms exactly what you and I have seen: That our performances have generally been fairly good. Many people look at our points tally and goal results to judge the performances, and conclude that we’ve been bad, but in our present context, xG sames more relevant than these stats. Because the context now is that this season we are trying to build a team and establish a new playing style, not actually trying to win the league (in ehich case losing points when performances are good would be a more crucial flaw).

Gaffes by players in new constallations are easier to iron out than if we constantly let opponents through on david, which we’ve seen a bit of in previous seasons. I take it as a good sign, nothing more.
I think it's more thna gaffes though. The lethargic first half against Palace was very concerning and the lack of any organisation or generally having a clue in certain situations (crosses, set pieces) is really very poor. There are positive elements in it too which ties in to what you are saying. The second half of the Chelsea game, first half at Wolves, and periods of the Southampton game, I think we've looked quite good. It's unrealistic to expect us to look good for 90 minutes of every game but we need to start using these periods in games to win them by not completing switching off in other parts or ignoring basic fundamentals.

Think I've changed my opinion a bit. at the time I said the Palace and Southampton performances were too similar to under Jose, but on reflection I don't think they were. We are just having different kinds of problems. If we can iron them out then at the very least it'll be much easier to get behind this team, even if the results are still a bit hit and miss. if some of the rest of the team can show somewhere near the work ethic of James that will also help.
 

chromepaxos

Full Member
Joined
Jul 29, 2015
Messages
184
xG is a good idea in theory, but it just doesn't work in football when sample sizes are so low and teams don't play each other enough times for the numbers to be relevant at all.
This is complete nonsense. xG is better predictor of points achieved than pretty much any other stat. The fact that you think it depends upon the number of times each team plays another shows that you're a bit clueless about how it works.
 

SAFMUTD

Full Member
Joined
Mar 14, 2018
Messages
1,888
I think people are misinterpreting the xG statistic, if we got a higher xG than goals scored it doesnt mean that magically our strikers will get more goals, its not luck its about ability.

Thats why Aguero always has more goals that xG expected, he is world class and thats what worldclass players do. They get goals in situations other strikers dont. At the end of the day the xG is a value given from many different players under many different chances but if you have a striker who misses a lot if wont matter if a thousand players would had scored, we will still struggle.

The same applies for our defence.
 

chromepaxos

Full Member
Joined
Jul 29, 2015
Messages
184
Do you know what can also give you a good reflection of teams' performance over long periods? Your eyeballs; your brain. Try using them. Some of you xG guys need to close your laptops, climb the stairs from your mother's basement and go have a beer, meet some people, get laid, do something, anything.
Because if there's one thing that the Caf proves, it's that everyone's subjective opinions are the same? That's frickin' hysterical.

Hey Ole! What you really need to do to analyse team performance is go down the pub and ask billybee99!

I mean, it's okay to be scared of maths, but pretty funny to sneer at people who aren't.
 

adexkola

Arsenal supporter
Joined
Mar 17, 2008
Messages
37,483
Location
The CL is a glorified FA Cup set to music
Supports
orderly disembarking on planes
Do you know what can also give you a good reflection of teams' performance over long periods? Your eyeballs; your brain. Try using them. Some of you xG guys need to close your laptops, climb the stairs from your mother's basement and go have a beer, meet some people, get laid, do something, anything.
:lol:
 

elmo

Can never have too many Eevees
Joined
Jul 14, 2008
Messages
7,365
Location
AKA: Slapanut Goat Smuggla
This is complete nonsense. xG is better predictor of points achieved than pretty much any other stat. The fact that you think it depends upon the number of times each team plays another shows that you're a bit clueless about how it works.
My problem with xG is sample size and inconsistency in the data being compared to come out with inaccurate conclusions.

You've people coming to conclusions based on stats to fit their own narrative without actually proving that their shit is even relevant and people are just lapping it up because it's much easier to blame it on bad luck than to admit that there's something really wrong with the fact that underperforming players aren't being held accountable.

We aren't unlucky so far this season, we've only played well for 60 mins max each game so far, hence our results.
 
Last edited:

SteveW

Full Member
Joined
Mar 6, 2013
Messages
6,663
You shouldnt need xg to know that we've been very unlucky so far. It's been blatantly obvious.
 

shabadu84

Mint? Berry?
Joined
Jan 12, 2010
Messages
4,606
Location
Muppet Treasure Island
Do you know what can also give you a good reflection of teams' performance over long periods? Your eyeballs; your brain. Try using them. Some of you xG guys need to close your laptops, climb the stairs from your mother's basement and go have a beer, meet some people, get laid, do something, anything.
Always amazed that people like this are able to find time to share responses like this in between the copious amounts of sex they're undoubtedly having.
 

Cassidy

No longer at risk of being mistaken for a Scouser
Joined
Oct 2, 2013
Messages
19,055
My problem with xG is sample size and inconsistency in the data being compared to come out with inaccurate conclusions.

You've people coming to conclusions based on stats to fit their own narrative without actually proving that their shit is even relevant and people are just lapping it up because it's much easier to blame it on bad luck than to admit that there's something really wrong with the fact that underperforming players aren't being held accountable.

We aren't unlucky so far this season, we've only played well for 60 mins max each game so far, hence our results.
Not playing well doesn't mean you have not been unlucky
 

bosnian_red

Worst scout to ever exist
Joined
Aug 13, 2011
Messages
37,221
Location
Canada
xG is a terrible stat for predicting future long term performance unless you believe that teams stay the same over the years despite multiple transfers and promotion/relegations.

xG is a good idea in theory, but it just doesn't work in football when sample sizes are so low and teams don't play each other enough times for the numbers to be relevant at all.
Not over years, but throughout the years for sure and between seasons if there are no drastic changes sure. Also useful to see how a team improves/declines between season as a result of transfers or a lack thereof. By long term I mean over the course of a season. The longer the season goes on, the more accurate the xG should get and it points to the general state of the team, if things are working, if they are broken, or if you're getting lucky or unlucky. Nobody is saying it's the be all and end all, but it's definitely a very useful stat and something that I would guarantee the vast majority of football clubs actually use in some way.