Thoughts v Cardiff

[Matt Stevenson]

We all get different things from watching football. Our perceptions and enjoyment vary from person to person depending on our age, how much we have played the game, what teams and players we have watched and the number of games attended, and what particular aspects of the beautiful game we find most fascinating. The loyalties handed down to us by our family members are hugely important too.
It is interesting that Matt S is so intrigued by statistics generated by games (and for me xG is a psuedo statistic based on someones's opinions and algorithms that can't possibly account for the thousands of variables that define each opportunity). Even the widely accepted possession percentage seems a bit bogus to me - for example, which team is in possession of the ball when the goalie kicks it 50 or 60 yards through the air from a goal kick, or there is a bout of head tennis in the centre circle, or during a goal mouth scramble when half the players on the pitch are trying to get at the ball? These percentages always add up to 100, but there really should be a third percentage to indicate how long the ball was in the air and/or when neither team could claim to have possession of it.
Personally, I go to football for the love of the game and the enjoyment of the spectacle, seeing great players in action, marvelling at great pieces of individual skill or the uncanny magic of fabulous team moves, memorable goals, fantastic goalkeeping, in a word the excitement of the contest.
There are magic moments that live in the memory that have less than zero to do with statistics, real or bogus: Gordon Banks saving from Pele in Mexico in 1970, Nobby Stiles with his toothless grin jogging round Wembley with the World Cup in 66, seeing George Best slalom through half the Saints team to score a wonder goal at the Dell, Bruce Grobbelaar flying through the air to pluck a goal bound shot out of the sky - the list is endless. Who cares what the xG was for Bretts solo goal in the 5-1 at Craven Cottage, or Steve Cook's piledriver in the same game. And what about that magic moment when Pughie broke the tension at DC like puncturing a huge balloon with the frst goal against Bolton that sent us into the PL for the first time?
And what statistics could ever replace the gale of laughter that engulfed the whole ground when Warren Aspinall playing at DC for Brighton, and who had been booed relentlessly from the start of the game, was poleaxed when hit in the nuts twice in the space of five minutes, gamely getting up to carry on and then receiving good natured cheers instead of boos when he finally left the field.
For me first and foremost watching football is an emotional experience, and it's not just the action. Who present could ever forget the wonderful reception Ian Cox received from our supporters when he returned to DC with Burnley only a few weeks after he left us, or the heartfelt applause given to the young boy who came on to the pitch on Monday to ring a bell in the centre circle to signify his recovery from cancer?
So, sorry Matt, for me football has way to much to offer the senses and the emotions to boil it down to a few often meaningless numbers. I don't knock your fascination with numbers, although I don't share it myself, but I hope that you are not missing all the things that I describe when you examine them.

I actually agree with these sentiments (so I'm clearly misrepresenting myself). What I am intrigued about though is why people say they are worthless / meaningless, but then make statements such as 'We should have been 2 up / 3 down, without realising what they are doing.

The xG timelines are just for casual interest, for me initially and then for other AFCB fans who might enjoy them. I think the real lashback came when they were used to suggest we were a midtable side which has never been my agenda.

 

[Matt Stevenson]

DJ, looking at this thread entitled Thoughts vs Cardiff I can't help feeling you should start a new one for XG and merge half the posts on here.

I was wondering whether to set up an xG2022 thread and just put those in there and that way only those who want to look / comment can do. I'll do that for the Peterborough game

 

[Neil Dawson]

Like all stats though it depends what you use them for. The fact that it doesn't allow for the player involved surely tells you about a striker's ability. If he's outscoring his xG he's clearly better than the average striker given the chances he's had. There's obviously a lot of focus on Solanke missing loads of sitters yet his xG would suggest he actually does better than most strikers (I think - someone may want to correct me). Same goes for the team as a whole.

If teams are creating high xG scored and low xG conceded and still aren't winning games it can point to where the problem is. If the opposite is true they are arguably relying too heavily on a decent striker or keeper.

Agree with all of that. It’s people saying we should be lower or higher in a league table that are ignoring the fact if you have a lethal striker you only need to create two chances in a game whereas an other team may need to create eight or nine. That’s not a predictor of where you will finish unless said striker gets injured.

 

[DJ]

I was wondering whether to set up an xG2022 thread and just put those in there and that way only those who want to look / comment can do. I'll do that for the Peterborough game

Sounds a good idea. Should create plenty of discussion game by game and be easy enough for those not interested to avoid.

 

[Matt Stevenson]

data is primarily the x, y location on the field where an attempt originates, whether the attempt is made with the foot, the head or another body part and a descriptive assessment of which type of phase of play preceded the attempt.

These numerical inputs are then used to calculate the likelihood that the attempt at scoring will be successful and the visual depiction of this information is uploaded

so, as I suspected, very limited; no mention of capability of the player having the chance, strong/weak foot, defensive cover, keeper position etc unless some of that is included in the visual depiction data.

Either way a very blunt tool

I'd say a blunt-ish tool . The position of the keeper / defenders is factored in though as is the flight of the ball. It's in this vid (How Did These Goals Go In? - We Explain How Goal Probability Works - YouTube )

Out of interest, NFC, which chance did you think was the easiest of the Anthony / Moore *2 ones? I couldn't call it, all were near enough to 50% and all were missed.

 

[Matt Stevenson]

Agree with all of that. It’s people saying we should be lower or higher in a league table that are ignoring the fact if you have a lethal striker you only need to create two chances in a game whereas an other team may need to create eight or nine. That’s not a predictor of where you will finish unless said striker gets injured.

Equally agree, the best predictions come from different sites (where currently on 538 we are predicted to finish in second, 6 points clear of Blackburn).

 

[SlowDownDerek]

We all get different things from watching football. Our perceptions and enjoyment vary from person to person depending on our age, how much we have played the game, what teams and players we have watched and the number of games attended, and what particular aspects of the beautiful game we find most fascinating. The loyalties handed down to us by our family members are hugely important too.
It is interesting that Matt S is so intrigued by statistics generated by games (and for me xG is a psuedo statistic based on someones's opinions and algorithms that can't possibly account for the thousands of variables that define each opportunity). Even the widely accepted possession percentage seems a bit bogus to me - for example, which team is in possession of the ball when the goalie kicks it 50 or 60 yards through the air from a goal kick, or there is a bout of head tennis in the centre circle, or during a goal mouth scramble when half the players on the pitch are trying to get at the ball? These percentages always add up to 100, but there really should be a third percentage to indicate how long the ball was in the air and/or when neither team could claim to have possession of it.
Personally, I go to football for the love of the game and the enjoyment of the spectacle, seeing great players in action, marvelling at great pieces of individual skill or the uncanny magic of fabulous team moves, memorable goals, fantastic goalkeeping, in a word the excitement of the contest.
There are magic moments that live in the memory that have less than zero to do with statistics, real or bogus: Gordon Banks saving from Pele in Mexico in 1970, Nobby Stiles with his toothless grin jogging round Wembley with the World Cup in 66, seeing George Best slalom through half the Saints team to score a wonder goal at the Dell, Bruce Grobbelaar flying through the air to pluck a goal bound shot out of the sky - the list is endless. Who cares what the xG was for Bretts solo goal in the 5-1 at Craven Cottage, or Steve Cook's piledriver in the same game. And what about that magic moment when Pughie broke the tension at DC like puncturing a huge balloon with the frst goal against Bolton that sent us into the PL for the first time?
And what statistics could ever replace the gale of laughter that engulfed the whole ground when Warren Aspinall playing at DC for Brighton, and who had been booed relentlessly from the start of the game, was poleaxed when hit in the nuts twice in the space of five minutes, gamely getting up to carry on and then receiving good natured cheers instead of boos when he finally left the field.
For me first and foremost watching football is an emotional experience, and it's not just the action. Who present could ever forget the wonderful reception Ian Cox received from our supporters when he returned to DC with Burnley only a few weeks after he left us, or the heartfelt applause given to the young boy who came on to the pitch on Monday to ring a bell in the centre circle to signify his recovery from cancer?
So, sorry Matt, for me football has way to much to offer the senses and the emotions to boil it down to a few often meaningless numbers. I don't knock your fascination with numbers, although I don't share it myself, but I hope that you are not missing all the things that I describe when you examine them.

This is a great summary of what's great about football but I really don't know what any of it has to do with statistical analysis after the event. Why on earth would xG take anything away from any of these things?

As has been stated ... whether anyone likes it or not, if you've walked away from a game saying "we should never have lost that" or "we should have scored three or four more" etc etc then you've performed a mental form of xG in your mind based on what you've seen. Everyone does it and they always have.

 

[LewSwimmin]

I actually agree with these sentiments (so I'm clearly misrepresenting myself). What I am intrigued about though is why people say they are worthless / meaningless, but then make statements such as 'We should have been 2 up / 3 down, without realising what they are doing.

The xG timelines are just for casual interest, for me initially and then for other AFCB fans who might enjoy them. I think the real lashback came when they were used to suggest we were a midtable side which has never been my agenda.

I think the point I was trying to make in mentioning the 5-1 Fulham game was that probably everyone who was there carries a memory of those two sensational goals, but I would be very surprised indeed if anyone has an equally clear recollection of our xG for the game!

 

[Jules]

data is primarily the x, y location on the field where an attempt originates, whether the attempt is made with the foot, the head or another body part and a descriptive assessment of which type of phase of play preceded the attempt.

These numerical inputs are then used to calculate the likelihood that the attempt at scoring will be successful and the visual depiction of this information is uploaded

so, as I suspected, very limited; no mention of capability of the player having the chance, strong/weak foot, defensive cover, keeper position etc unless some of that is included in the visual depiction data.

Either way a very blunt tool

What about factoring in the teams strengths, position in table etc?

 

[blandford_cherry]

But Neil / others, are xG stats more meaningless as clear chances / chances / red cards / penalties / possession etc etc ? If people only want one stat, the scoreline, (which is ultimately the only thing that matters for that one game) they should equally disregard anything else. I'm bemused as to why xG get the stick whereas other stats don't, so I am interested in people justifying the distinction.

I suspect it is because they were used out of context to predict that AFCB weren't that good, and as Cherry fans we were never going to like that. The underlying principle of that thread which was 'if we stop being as clinical, and the opposition start being more clinical we won't do as well', is hard to dispute, but is not the best foundation for predictions with different quality players.

For future predictions, that do consider how many points have been won to date I prefer 538. Encouraging predictions are below.

View attachment 6983

Responding to other comments, we were unlucky not to be in front against Middlesbrough at half time (it was the second half where we didn't turn up). The xG was 0.67 vs 1.51. The Christie miss still makes me wince.

I'll put the timeline up shortly, but it resembles the comments in the match thread that Cardiff had two big chances before we really got going, and then we were in control.

What is the website that the screenshot is taken from? Looked at it quite often last season but completely forgot the address?

 

[wingsnwheels]

How can stats account for things like the event in the late 1970's early 80's when Keith Miller gave away a penalty. Our keeper (Kenny Allen?) had the ball in his arms and Miller walked up to the keeper and spoke to him. He appeared to have poked the keeper on his chest and the lino (assistant) flagged for hand ball saying that Miller's arm touched the ball. Would that be given nowadays as it was not a deliberate act?

 

[Matt Stevenson]

What is the website that the screenshot is taken from? Looked at it quite often last season but completely forgot the address?

https://projects.fivethirtyeight.com/soccer-predictions/championship/

 

[Waz afcb]

data is primarily the x, y location on the field where an attempt originates, whether the attempt is made with the foot, the head or another body part and a descriptive assessment of which type of phase of play preceded the attempt.

These numerical inputs are then used to calculate the likelihood that the attempt at scoring will be successful and the visual depiction of this information is uploaded

so, as I suspected, very limited; no mention of capability of the player having the chance, strong/weak foot, defensive cover, keeper position etc unless some of that is included in the visual depiction data.

Either way a very blunt tool

I’m aware that this angry man has blocked me but just to add to the debate, why does XG receive so much criticism when other stats do not? I can guarantee you if we played Man City and had 20 shots on target compared to their one but ended up losing one nil, the overwhelming feeling would be ‘we were the better team and deserved to win that game,’ not ‘Man City deserved to win because their more talented striker converted their only opportunity.’ I feel like XG is as legitimate stat as any and that anybody suggesting otherwise is doing so to try and make out that we are a better team than we actually are.

 

[nd9cherry]

I was wondering whether to set up an xG2022 thread and just put those in there and that way only those who want to look / comment can do. I'll do that for the Peterborough game

Never really got this football thingy...56 years supporting the cherries and the announcer finally tells me what the half time and full time score was!!!!

 

[Waz afcb]

This is a great summary of what's great about football but I really don't know what any of it has to do with statistical analysis after the event. Why on earth would xG take anything away from any of these things?

As has been stated ... whether anyone likes it or not, if you've walked away from a game saying "we should never have lost that" or "we should have scored three or four more" etc etc then you've performed a mental form of xG in your mind based on what you've seen. Everyone does it and they always have.

Very well put.

 

[CherryBare]

This is a great summary of what's great about football but I really don't know what any of it has to do with statistical analysis after the event. Why on earth would xG take anything away from any of these things?

As has been stated ... whether anyone likes it or not, if you've walked away from a game saying "we should never have lost that" or "we should have scored three or four more" etc etc then you've performed a mental form of xG in your mind based on what you've seen. Everyone does it and they always have.

Yes maybe, but I've never said, oh we're in a false position in the league, because over the season our XG is lower than the goals we scored

 

[kirsikka]

But you may have said 'If such and such had taken the great chances we created but he missed then we'd be much better off in the league' which is the point SDD is making.

 

[zigger zaggar]

Confession time: Although watching when that challenge on Billing went in, I completely didn't see how bad it was. I was really surprised when a scrap ensued and even more when the ref brandished his red card. My eyesight is failing!

Glad to see waz is back with his trolling negativity.

Is it not possible to troll with positivity?
Only asking

 

[New Forest Cherry]

I'd say a blunt-ish tool . The position of the keeper / defenders is factored in though as is the flight of the ball. It's in this vid (How Did These Goals Go In? - We Explain How Goal Probability Works - YouTube )

Out of interest, NFC, which chance did you think was the easiest of the Anthony / Moore *2 ones? I couldn't call it, all were near enough to 50% and all were missed.

good question, although I was questioning the mechanics of the algorithm not the results
if Lewandowski had been presented with those chances would the XG output have been different?

 

[SlowDownDerek]

Yes maybe, but I've never said, oh we're in a false position in the league, because over the season our XG is lower than the goals we scored

No, that in my opinion is stretching the usefulness of the statistic. There's too many flaws in it to be a good indicator of what the outcome of games should be.

People have mentioned things like great chances that never actually resulted in a shot not being counted. Also a series of shots adds up cumulatively yet of the first effort goes in the others don't happen.

One other issue is that goals change games and if you're winning you don't need to try and score, especially in tight away games. Your xG is therefore affected by the scoreline.