Author Topic: 'It's beyond the model': have Liverpool exposed the limits of xG?  (Read 10392 times)

Offline Alan_X

  • WUM. 'twatito' - The Cat Herding Firm But Fair Voice Of Reason (Except when he's got a plank up his arse). Gimme some skin, priest! Has a general dislike for Elijah Wood. Clearly cannot fill even a thong! RAWK Resident Muppet. Has a crush o
  • RAWK Staff
  • Legacy Fan
  • ******
  • Posts: 53,376
  • Come on you fucking red men!!!
  • Super Title: This is super!
Re: 'It's beyond the model': have Liverpool exposed the limits of xG?
« Reply #80 on: August 13, 2020, 06:15:50 pm »
You'd think we were by the way xG is waxed lyrical about in some quarters.


You can patronise all you want mate with sly comments, it doesn't phase me. I don't claim to be great with numbers and I won't be made to feel bad about that. I've just watched and played football for a hell of a long time to know that no matter what, data or numbers cannot replicate the unpredictability of a game. But if that isn't what xG promises then fine, I've already asked what it does promise. And if it doesn't promise all that much, then I don't see the clamour for something that doesn't even have a position.

I think there are two things going on. On the one hand there is the pundit and layman's use of things like xG to give another talking point about the game. It's a bit like the batting figures in baseball. They give a broad brush view of a player or a team's performance.

The other is the far more complex analysis that the club uses which is combined with intiuition and creativity of the manager and the coaching staff. I think the LFC analytics team has been a revelation for Klopp. He knows the kind of things he wants his team to do and the analytics team can show how that can be made tor work or how the players can be coached to see the runs, the moves ito space, the triggers for the press etc.

It's about fine margins and every small advantage.

One thing I don't agree with is the suggestion that 'we can't keep performing at this level - we have to revert to the mean at some point...' The point of analytics as carried out by our guys is to ensure those advantages continue by understanding them and replicating them or improving them.
Sid Lowe (@sidlowe)
09/03/2011 08:04
Give a man a mask and he will tell the truth, Give a man a user name and he will act like a total twat.
Its all about winning shiny things.

Offline newterp

  • RAWK Supporter
  • Legacy Fan
  • ******
  • Posts: 26,769
Re: 'It's beyond the model': have Liverpool exposed the limits of xG?
« Reply #81 on: August 13, 2020, 06:33:33 pm »
Now we just need and appearance by FazeofPlay and this thread can be closed.

Offline Dave McCoy

  • "Don't you know who I am?!"
  • Legacy Fan
  • ******
  • Posts: 7,461
  • 11,053ft up
Re: 'It's beyond the model': have Liverpool exposed the limits of xG?
« Reply #82 on: August 13, 2020, 06:41:06 pm »
But xG shouldn't be used on a game to game basis to make decisions.  It can help inform your opinion of what your eyes saw in a particular game but it's most accurate use is in long term trend lines on overall team performance.  There's a reason betting firms quickly embraced it to help take more of your money as it's pretty accurate in seeing what is "luck" and what is just a bad team in a run of results.  Where do you think Ted Knutson came from?  Not some idyllic football lab that's for sure.

As far as LFC, we already have the points and the titles.  Nothing changes that.  It's more a question of what should be the expectation for 20/21 be?  I'd say the problem here is that we clearly dropped a level the last 8 games, especially defensively, so any trend line is going to be unfairly skewed.  Then the 20/21 season is shaping up to be a hellish grind for the players, is obtaining the previous level even possible in those circumstances?  I would say that getting a bench that can contribute more consistently on the offensive end should be a priority as when you look at the numbers we really have no other attackers you'd count on outside the main 3.  If there is any drop off it's going to come from there.

Offline Hazell

  • Ultimate Movie Night Draft Winner 2017. King - or Queen - of Mystery. Hyzenthlay. The 5th Benitle's sex conch.
  • RAWK Scribe
  • Legacy Fan
  • ******
  • Posts: 76,836
Re: 'It's beyond the model': have Liverpool exposed the limits of xG?
« Reply #83 on: August 13, 2020, 06:51:30 pm »
What she said.

Xg is a static indicator based on averages and I would be surprised if it is used in a meaningful way by any top teams.  Shots in the box are more likely to score goals than those from out wide or further away? No shit.

The video link that Roy mentioned a few days ago from the Royal Institution Christmas Lectures highlights just how far beyond xg our stats guys are.The stats guys have created a 15 second moving window for goal probabilities from anywhere on the pitch (see below) - in this sense it is like a dynamic version of xg, but one that includes movements and passes most likely to lead to a goal, not just the point where the final shot is taken.  The model accounts for things like players' maximum speeds, but it wouldn't be at all surprising, for instance, if the maximum speed data is also dynamic i.e. it considers how many minutes the player has been on the pitch, how long they since they recovered from the last sprint etc.



Distilling multi-dimensional data into something comprehesible and meaningful like this animated 2D contour/heat map is an art form.  Techniques like Principal Component Analysis (PCA) can be used to reduced dimensionality without throwing information away and allow machine learning (AI) techniques to be used in a way that increases computational efficiency.  Outcomes from all conceivable situations can then be calculated in a timely manner to provide feedback to coaches and players.

Xg is fun for forum discussions and for pundits to fill column inches and the good ones who understand the metric can produce interesting analyses.  Although when the games are being used to discuss the model rather than the other way round it does start to get a bit meta.

Good post. I saw that Royal Institution Christmas Lecture and it was a great example of the kind of stuff the club looks at at. That's probably the tip of the iceberg as well, none of us would be able to understand the depth the analysts at the club go into. No wonder they employ people with PhD's in Theoretical Physics and the like.

The article in the OP is good as well. xG as a definitive argument is pointless.
We have to change from doubter to believer. Now.

Offline Red Bird

  • RAWK Supporter
  • Legacy Fan
  • ******
  • Posts: 3,250
  • We all Live in a Red and White Kop
Re: 'It's beyond the model': have Liverpool exposed the limits of xG?
« Reply #84 on: August 13, 2020, 08:14:33 pm »
This is it really. Expected goals in isolation can easily be misconstrued and doesn’t paint the whole picture. A week or two ago Dan Kennett posted a thread comparing our goal difference to City’s depending on whether we were winning, drawing, or losing. It’s not a surprise that we were equal or better than City when we were drawing or losing and they blew us away when they were winning. It’s obvious from watching our games that once we take the lead it is all about control. We rarely, if ever, run up the score on our opponents. Here’s the thread for those interested:

https://twitter.com/dankennett/status/1288450451671781376?s=21

Hole in one.

Offline JC the Messiah

  • Legacy Fan
  • ******
  • Posts: 3,629
  • ♪ ...and now Jürgen-a believe us... ♬
Re: 'It's beyond the model': have Liverpool exposed the limits of xG?
« Reply #85 on: August 13, 2020, 09:29:12 pm »
Weather forecasts.

You hear people saying that the weather forecast said rain, and it didn't rain where they were. Their conclusion, is that the weather forecast is rubbish, and therefore the metrologists are rubbish.

What they fail to take into account is:

1. The weather forecast they see is a summary of the data. They have a whole range of probabilities of rain, temperatures, etc., and need to summarise this with one symbol or number.

2. The forecast of for a wide area, we all know it can rain on one side of town whilst the other side is bathed in glorious sunshine.

3. It's a forecast, and there are so many factors that need to be accounted for, including the random event, such as the butterfly flapping its wings...

If football was easy, every manager would be doing what Klopp is doing, and Klopp wouldn't be able to command the salary he earns.
"I said to the boys before the game it would be impossible. But because it’s you, I say we have a chance."
Jürgen Klopp, 7 May 2019

"I told them if we score it will be different. We scored. It was different."
Rafael Benitez, 25 May 2005

Offline Skeeve

  • RAWK Supporter
  • Legacy Fan
  • ******
  • Posts: 15,792
Re: 'It's beyond the model': have Liverpool exposed the limits of xG?
« Reply #86 on: August 14, 2020, 01:39:20 pm »
The other factor is the game management angle. That we start fast, score and then play the game in a more controlled manner, not necesarily pushing for more xG. I think we just need a larger sample size to see if that really can play as big of a factor as some claim it does. Because I'm sceptical. Firstly, in all fields of human endeavour, we tend to underestimate just how much randomness is at play. We look at very small data sets and have an urge to make narrative-based assumptions before we see the longer term.

Secondly, it supposes that Pep Guardiola hasn't figured this out? That Klopp has some huge edge in his thinking that allows his teams to do this and Pep can't see it. And even further, that we can all see it, and Pep can't. As much as I'd love that to be true, I'm finding pretty hard to swallow. Considering City's points totals in their previous two seasons, it seem absurd to me.

So yeah, to my eyes, it looks like a season where we ran extremely well, and City absolutely did not.

The obvious point to consider with Guardiola is that his methods work for him and since he has so many years at clubs where budgets weren't an issue, he simply isn't going to need to consider such things to the same extent, he can just bring on a few more £50m players during the game instead.

We also shouldn't underestimate the work done by those around the managers, compare our success rate with signings to the other big pl sides in recent years, significantly better because we have a coherent plan that goes beyond just the manager himself.

Offline Skeeve

  • RAWK Supporter
  • Legacy Fan
  • ******
  • Posts: 15,792
Re: 'It's beyond the model': have Liverpool exposed the limits of xG?
« Reply #87 on: August 14, 2020, 01:45:16 pm »
Surely an easy fix would be to take in to account the quality of the person taking the shot and how good they are from that particular angle? Obviously somebody like Mane is gonna have a higher chance of scoring in the same situation than a lower level PL finisher. I believe Vardy is the PL player who most overachieves in terms of Xg to actual goals, which makes sense since he plays for a team that creates fewer chances and he has to be very clinical. Xg judges the quality of playmakers but doesn't account for better finishes.

That would add another level of complexity to the measure and more crucially would also be introducing a more subjective measure into it as well, like all stats, you need to understand their limitations and which ones are appropriate to use at which time rather than looking for a single easy option like tv has with xG.

Offline royhendo

  • RAWK Staff
  • Legacy Fan
  • ******
  • Posts: 253,074
  • We all Live in a Red and White Kop
Re: 'It's beyond the model': have Liverpool exposed the limits of xG?
« Reply #88 on: August 14, 2020, 02:28:34 pm »
Stop over-analysing the game. It is already watered down and not what it was in the past. Football is too unpredictable to be predictable.

This is possibly my favourite post of all time.

If people didn't over-analyse, we wouldn't have the computers/phones/tablets we're using now, the networks that run them and allow you to post, the screens that display the football matches you watch... you can take it all the way down to why they decided to have the penalty spot 12 yards from the goal, and the dimensions of the goal itself.

P.S. The reason we have our manager is that the analytics team proved his perceived drop off in performance at Dortmund was in fact just an incredibly unusual run of bad luck. Ian Graham's appearance on the Freakonomics podcast tells the story - he analysed ten seasons of league data and the Dortmund season was a freak one - in XG terms they were 2nd best in the league, but the team were in the relegation zone for much of the season.
« Last Edit: August 14, 2020, 02:35:31 pm by royhendo »
"Word of the day is 'philodox' (17th century): one who is in love with their own opinion, and who consequently believes that everyone else should share it."  @susie_dent on twitter - https://twitter.com/susie_dent/status/1419683653844668422

Offline Samie

  • The next Pharaoh of Egypt. The Ev of drafting! Rumoured to be the 7th, we may need that old magic back! The Timekeeper, ask him what time the action starts.
  • RAWK Supporter
  • Legacy Fan
  • ******
  • Posts: 66,424
Re: 'It's beyond the model': have Liverpool exposed the limits of xG?
« Reply #89 on: August 14, 2020, 02:33:24 pm »
Roy while I get your point about that part of Fiasco's post look at the rest of it. Analytics can give you a base  but in game situations are a diffrent ball game.  :D

Offline royhendo

  • RAWK Staff
  • Legacy Fan
  • ******
  • Posts: 253,074
  • We all Live in a Red and White Kop
Re: 'It's beyond the model': have Liverpool exposed the limits of xG?
« Reply #90 on: August 14, 2020, 02:47:22 pm »
Absolutely - the stats guys can probably do a bit better than a cumulative xg as well, they might, for example, integrate over the size of high density areas created by Bobby compared to those a lesser mortal might have generated in a similar position to determine additional effects.

This is exactly it. The understanding and ability to measure the right things improves. Over ten years we've gone from talking about disruptive activity from forwards (tackles, fouls, interceptions - all available from the Opta stuff that was about then), through to passes attempted/completed and passes received in the final third, through to packing... it stands to reason that the analysis will eventually find ways to normalise for key variables - certain measures got possession adjusted... then phases of play got factored in... now we have open talk of factoring in the height of the ball when a shot's taken on the XG for the attempt.

The players who are 'multipliers' (as PoP I think used to call them) effectively increase the size of the red areas.  Stats related to pass completion given the position of opposing players could easily provide such additional dimensions to the model which would show how indiviual players to extend and darken the red bits for their team mates.

Yeah - spotting an Alonso by the attributes of support running and staying 'open'. You could measure which players hide doing the same thing.

You can also argue that being surrounded by quick players gives a larger red target area as the red bits are the places where we can get too quicker than the opposition.  How often do we see a ball pinged up to Sadio or Mo and think 'he's never going to get to that' only to be pleasantly surprised.

And the speed of the carrier too (Luis Alberto v Adama Traore).
"Word of the day is 'philodox' (17th century): one who is in love with their own opinion, and who consequently believes that everyone else should share it."  @susie_dent on twitter - https://twitter.com/susie_dent/status/1419683653844668422

Offline Fiasco

  • Just add water to foam at the mouth. Can't spell San Francisco. Has promised to eat his own cock. Cannibal Self-Harm in that case.....
  • RAWK Supporter
  • Legacy Fan
  • ******
  • Posts: 22,257
  • JFT96.
Re: 'It's beyond the model': have Liverpool exposed the limits of xG?
« Reply #91 on: August 14, 2020, 04:00:21 pm »
This is possibly my favourite post of all time.

If people didn't over-analyse, we wouldn't have the computers/phones/tablets we're using now, the networks that run them and allow you to post, the screens that display the football matches you watch... you can take it all the way down to why they decided to have the penalty spot 12 yards from the goal, and the dimensions of the goal itself.

P.S. The reason we have our manager is that the analytics team proved his perceived drop off in performance at Dortmund was in fact just an incredibly unusual run of bad luck. Ian Graham's appearance on the Freakonomics podcast tells the story - he analysed ten seasons of league data and the Dortmund season was a freak one - in XG terms they were 2nd best in the league, but the team were in the relegation zone for much of the season.

How were teams ever successful before the age of technology then? You make it sound like football can only exist and get better because of it. That just isn't the case. I've already said that analytics and stats are more than useful, I'm not taking that away from it. I'm saying people are going way too far the other way.

Technology and phones? What does that have to do with whether a striker is likely to score when presented with a golden opportunity in a key match in the 90th minute? I'll say it again, go and watch or partake in a game at any level and you'll see a shedload and statistical quirks and abnormalities in just a short space of time.  People want to over-analyse a game that has so much of a human component to it and there is a limit of what you can do with that. I'll be told I'm wrong, but make the game robotic all you want, it isn't supposed to be that way.

You say xG told our stats gurus that Dortmund should have been second but weren't. Fair enough. But how many could see that anyway? I remember around that time many on here and other outlets were saying the same thing and from the Dortmund games I watched myself it was clear that they were much better than their table showed. So why give xG so much credit for pointing that out? Anyone with half a footballing brain could see that and therein lies my beef for want of a better word: you want to give xG credit for being ahead of the curve and knowing that Dortmund were hard done by and should've been higher up the table, but human beings with eyes and feelings and self-analysis could see that.

I'll reiterate once again, stats, computers, models and all of that have their place. Don't get me wrong. I'm saying that too much credit is given to them and before long we might as well run a simulation of a game and base the result on that, because at the the end of the day, that simulation result would be the real result.


Offline Samie

  • The next Pharaoh of Egypt. The Ev of drafting! Rumoured to be the 7th, we may need that old magic back! The Timekeeper, ask him what time the action starts.
  • RAWK Supporter
  • Legacy Fan
  • ******
  • Posts: 66,424
Re: 'It's beyond the model': have Liverpool exposed the limits of xG?
« Reply #92 on: August 14, 2020, 04:18:20 pm »
Okay 2 examples say from this season:

Scenario A- Sadio Mane on goal one-on-one against say Nick Pope and he's scored 8 goals in the last 8 games, xG tells you he's going to score right? He probably scores anyway.  :D

Scenario B- Sadio Mane on goal one-on-one against Nick Pope again and this time he's not scored in 8 games.So does xG tell me he's going to score this or not? The doubt of this Sadio in his mind because he's not score in 8 might just give Pope the chance to save it or Sadio fucks up.

So in these 2 situations a confident Sadio and not a confident Sadio, XG can't predict that.

Online Chris~

  • Legacy Fan
  • ******
  • Posts: 12,562
  • We all Live in a Red and White Kop
Re: 'It's beyond the model': have Liverpool exposed the limits of xG?
« Reply #93 on: August 14, 2020, 04:22:17 pm »
Okay 2 examples say from this season:

Scenario A- Sadio Mane on goal one-on-one against say Nick Pope and he's scored 8 goals in the last 8 games, xG tells you he's going to score right? He probably scores anyway.  :D

Scenario B- Sadio Mane on goal one-on-one against Nick Pope again and this time he's not scored in 8 games.So does xG tell me he's going to score this or not? The doubt of this Sadio in his mind because he's not score in 8 might just give Pope the chance to save it or Sadio fucks up.

So in these 2 situations a confident Sadio and not a confident Sadio, XG can't predict that.
It's not giving you a yes/no prediction. Thats just a complete misunderstanding of expected goals.

Offline Ghost Town

  • RAWK snitch. Bands won't play no more. Too much fighting on the dance floor! Probably one of only three people who knows the meaning of "depuratory", the Suzy Dent-esque freakshow! Hannoying homunculus.
  • Legacy Fan
  • ******
  • Posts: 14,899
  • mundus vult decipi
Re: 'It's beyond the model': have Liverpool exposed the limits of xG?
« Reply #94 on: August 14, 2020, 04:31:53 pm »
Okay 2 examples say from this season:

Scenario A- Sadio Mane on goal one-on-one against say Nick Pope and he's scored 8 goals in the last 8 games, xG tells you he's going to score right? He probably scores anyway.  :D

Scenario B- Sadio Mane on goal one-on-one against Nick Pope again and this time he's not scored in 8 games.So does xG tell me he's going to score this or not? The doubt of this Sadio in his mind because he's not score in 8 might just give Pope the chance to save it or Sadio fucks up.

So in these 2 situations a confident Sadio and not a confident Sadio, XG can't predict that.
xG never tells you whether a player is going to score or not, so you're on the wrong track right from the off.

xG will give you an idea of what the likelihood is that players in the same/very similar position have scored in the past, according to all the historical data they have. This will give you an idea of what the mode outcome has been in the past. As an analyst you can then compare that to what actually happens in this instance and thus get some interesting data. The data might be that Mane's outcome in this instance is as predicted, from which something can be learned.

Or it might show that something different has happened, in which case something can also be learned. In fact there's a good case to be made that there's more to be learned - or least more data worth analysing - when the outcome in this instance varies significantly from the xG. Why does it? Are we doing something better/worse that we can build on or rectify? Is our player better/worse? Does he need extra training in some skill? Or did the opposition react in a better/worse way and can we learn how to do things differently to gain an advantage? And so on.

Similarly if at the end of the game the xG sum echos the actual result there's data there. And if it doesn't; if the result is contrary to the xG there's data there as well. Things to analyse and learn from.
"Every man has a right to utter what he thinks truth, and every other man has a right to knock him down for it."
Samuel (not Glen) Johnson, as reported by James (not Joey) Boswell. They must have foreseen RAWK ;D

Offline Samie

  • The next Pharaoh of Egypt. The Ev of drafting! Rumoured to be the 7th, we may need that old magic back! The Timekeeper, ask him what time the action starts.
  • RAWK Supporter
  • Legacy Fan
  • ******
  • Posts: 66,424
Re: 'It's beyond the model': have Liverpool exposed the limits of xG?
« Reply #95 on: August 14, 2020, 04:34:30 pm »
Yeah i get it's not Yes/No answer but  Sadio in both instances would say he should score right? But one is a confident Sadio and the other one isn't. But in both scenarios he might score, he might not score or one or the other.  :D

Offline Ghost Town

  • RAWK snitch. Bands won't play no more. Too much fighting on the dance floor! Probably one of only three people who knows the meaning of "depuratory", the Suzy Dent-esque freakshow! Hannoying homunculus.
  • Legacy Fan
  • ******
  • Posts: 14,899
  • mundus vult decipi
Re: 'It's beyond the model': have Liverpool exposed the limits of xG?
« Reply #96 on: August 14, 2020, 04:40:21 pm »
Yeah i get it's not Yes/No answer but  Sadio in both instances would say he should score right? But one is a confident Sadio and the other one isn't. But in both scenarios he might score, he might not score or one or the other.  :D
Sure, so if he doesn't score when xG suggests the mode outcome is that a good striker should, then we can analyse why. Is it confidence? Is he tired? Upset about something? Are his boots too tight? Is his positioning or technique marginally off?

Sure sometimes the 'eye test' will warn you that something is up with a player, but other times it's not as easy to tell. The data might be the first indication that all is not well and worth investigating and then rectifying.
"Every man has a right to utter what he thinks truth, and every other man has a right to knock him down for it."
Samuel (not Glen) Johnson, as reported by James (not Joey) Boswell. They must have foreseen RAWK ;D

Offline royhendo

  • RAWK Staff
  • Legacy Fan
  • ******
  • Posts: 253,074
  • We all Live in a Red and White Kop
Re: 'It's beyond the model': have Liverpool exposed the limits of xG?
« Reply #97 on: August 14, 2020, 04:44:15 pm »
How were teams ever successful before the age of technology then?

The age of technology precedes the age of football, Fiasco. Sides break the mould because analysis allows them to out-think their opponents. That goes back as far as competition does, but specifically, because it's relevant, I tend to think of Jock Stein sitting in the dugout drawing on pitch diagrams to figure out where the other side's play gets directed through, or Shanks building his sweat box, or Michels figuring out that zonal overloads combined with multi-functional players gives his teams a massive advantage.

Let's get the chicken and egg element of this out of the way before we go on.

Football exists full stop - it's independent of the analysis*
Analysis exists as a derivative - it's intertwined with the subject, whatever the subject is.
Football gets refined as a result of that analysis.

Or to rephrase for the geeks...

10. MAKE FOOTBALL
20. ANALYSE FOOTBALL
30. GOTO 10

You make it sound like football can only exist and get better because of it. That just isn't the case. I've already said that analytics and stats are more than useful, I'm not taking that away from it. I'm saying people are going way too far the other way.

Now - let's go back a bit. Here's where you came in.

One of my main issues with xG is that it doesn't take into account the human aspect of the game ...football isn't an exact science and especially not when you look at things in the long term.

But oddly, you then, despite saying it's not an exact science, define a number of attributes that might be useful to measure, and therefore improve the precision of the science.

...does anything else come into it? Who the goalie is? The angle? The form of the striker? The conditions? Whether his team is 4-0 up or whether it is goalless and in the last minute? Did the striker scuff the shot? Was a defender closing him down as he hit it? Was he wary of being offside and yet not fully committed? The possibility of the goalie making a world class save?

I could ask 10 other questions like that.

Yes, and in doing so, you'd be doing analysis (or proposing things that might potentially improve it).

What happens on a pitch happens on a pitch because of 1000s of variables.

Exactly. More and more of them are measurable, too.

Stop over-analysing the game. It is already watered down and not what it was in the past. Football is too unpredictable to be predictable.


There's then a series of posts where people who like stats say things like 'the things you say are common in people who are sceptical about stats, which is annoying, and then you say things like this.

...is it so otherworldly that a luddite like me could never understand it?

You can patronise all you want mate with sly comments, it doesn't phase me.

Now - can we all stop that?

Nobody seems to be claiming that stats are an end in themselves. The stats that matter are goals for and against, and points. That's about it. Everything else has to contribute to those ends, be they glamorous German nutritionists, Portuguese youth coaches, PHDs in astrophysics, or whatever else.

I don't find Liverpool's football robotic. I find Man City's football can get that way as much as I can with Simeone's sides. We kind of go another way and kind of harness chaos in our approach, despite our side having probably the most direct analytical input of any set up since Charles Reep's bollocks in the middle of the last century that led all the way to the Crazy Gang.

You say xG told our stats gurus that Dortmund should have been second but weren't. Fair enough. But how many could see that anyway? I remember around that time many on here and other outlets were saying the same thing and from the Dortmund games I watched myself it was clear that they were much better than their table showed. So why give xG so much credit for pointing that out?

Cos the fella marched through to the men in the suits and said, "Klopp's stock is low - maybe we can get him", then showed him his workings, and they acted on it.

It seems to me that what you're suggesting is pretty simple: that analysis, in and of itself, waters down the beauty of the game. That magic (the wonder of not understanding how things work) will be diminished to zero if people think too much. Me personally, I think that says more about you than it does the limits of the magic of the game. The beauty of the game is boundless - it can never be over-analysed. You can go from noughts and crosses, to stone/paper/scissors, to Connect 4, to whatever else through to football. Even the simplest of games are pretty straightforward to analyse, but somehow people keep playing and enjoying them. The more variables there are, the more infinitely complex the game can be, and the less it yields to being 'perfected'. The minute you go with a certain approach, someone finds a way to work around it.

Football is just about the richest game there is, in terms of its complexity.
« Last Edit: August 14, 2020, 05:31:38 pm by royhendo »
"Word of the day is 'philodox' (17th century): one who is in love with their own opinion, and who consequently believes that everyone else should share it."  @susie_dent on twitter - https://twitter.com/susie_dent/status/1419683653844668422

Offline Fiasco

  • Just add water to foam at the mouth. Can't spell San Francisco. Has promised to eat his own cock. Cannibal Self-Harm in that case.....
  • RAWK Supporter
  • Legacy Fan
  • ******
  • Posts: 22,257
  • JFT96.
Re: 'It's beyond the model': have Liverpool exposed the limits of xG?
« Reply #98 on: August 14, 2020, 05:27:54 pm »
Roy, I defined things that you claim help to make it an exact science as examples as to why it isn't, we're just looking at that differently. You reckon because there are so many variables that you can pinpoint those variables and chip away at them until you get a grasp of it and then get an advantage? My argument is counter to that, my argument is that in a game of football there are so many variables and bizarre events that you can't ever get a grasp on them fully. You are talking in the billions perhaps when you take in a whole season. I don't think you can ever fully get a handle on that. Nature is too unpredictable taking away the footballing aspect, the bounce of the ball, the ricochet, the angles, the reaction times, I mean everything. And it isn't the same from one player to the next, it isn't even the same from the one player hitting the same type of shot 2 minutes apart. You are talking along the lines of infinite universe territory here.
 
You're right in what you say about my opinion being that over-analysis waters down the beauty of the game. That is my view. Before lockdown I went to watch my mate play in a semi-professional match. A relatively decent standard of players, many had been at good clubs but never quite made it. Anyway, the game ebbed and flowed and for a feisty encounter it was full of good quality. It finished 4-3 and nobody could take their eyes of it. Again, the managers were making their way in the game and it wasn't a kickabout, you could see they were well coached and knew their roles. But at no point did any sort of analysis come through. Both sides missed chances, one lad scored a worldie and another missed golden opportunities. It was the best game of live football outside of a Liverpool game I'd seen in my life. At no point did any analysis come into it about xG, about key passes, about how many successful passes a certain player made to a certain teammate at a certain point that might or might not have created a chance. The chat was about a proper good game of football between two good sides played the right way.

Sure, the managers and coaches afterward would analyse it but after that I couldn't give a fuck about the stats, I watched a boss game of football. Now your argument will be about nice guys finishing last and how if you want to win you have to look beyond the spectacle and get to the nitty gritty. Noted, and I don't disagree. Yet every single kid out there now in a park aspiring to be the next Messi will be playing football for the beauty of it. He or she will fall in love with it because of the joy, the passion, the frustration it gives, the beauty it gives. Football exists in that way in spite of stats. Stats, xG and the like didn't invent the game of football. Football is a world unto itself and for all I'll be mocked and be told my backward thinking doesn't get it, I'm fine with that too. But what exactly does that say about me Roy like you mention above?

If results in football were universally fair then we wouldn't have Istanbul, we wouldn't have number 6, we wouldn't have Gerrard's FA Cup final in 2006, we wouldn't even be champions now according to one model. Leicester would never have been champions, because we were told throughout the season that what they were doing was unsustainable. I could go on and on and on. On the flipside, the data you trust has probably called quite a few things right too. Which lends me to think that the truth is probably somewhere in the middle as it often is.



Offline afc tukrish

  • How long for them sausages? Maggie May's Mythical Turkish Delight. RAWK's Expert Sausage Monster! Oakley Cannonier is fucking boss. Likes blowing his friends and undoing their nuts? Who nose?!
  • RAWK Supporter
  • Legacy Fan
  • ******
  • Posts: 16,933
  • This looks like a nice spot...
    • Flat Back Four
Re: 'It's beyond the model': have Liverpool exposed the limits of xG?
« Reply #99 on: August 14, 2020, 05:29:15 pm »
Yeah i get it's not Yes/No answer but  Sadio in both instances would say he should score right? But one is a confident Sadio and the other one isn't. But in both scenarios he might score, he might not score or one or the other.  :D

Measure confidence.

Can't be done, can it?

Doesn't have a place in this discussion, it's a peculiarly Brit obsession entirely subjective in nature...

If a confident Sadio doesn't score, was he not confident in the first place? And if an un-confident Sadio does finish, was he not confident in the first place?

Or is the whole idea not subjectively laid onto events after the fact?
Since haste quite Schorsch, but Liverpool are genuine fight pigs...

Offline El Lobo

  • Chief Suck Up. Feel his breath on your face. Toxic, pathetic, arse-faced, weaselling slimeball. RAWK Maths Genius 2022.
  • RAWK Supporter
  • Legacy Fan
  • ******
  • Posts: 54,990
  • Pretty, pretty, pretty pretty good
Re: 'It's beyond the model': have Liverpool exposed the limits of xG?
« Reply #100 on: August 14, 2020, 05:32:58 pm »
If a Confident Mane leaves Liverpool at 3PM travelling at 60mph and an Unconfident Mane leaves Glasgow at 5PM travelling at 90MPH, does the Confident Mane make a sound?
If he's being asked to head the ball too frequently - which isn't exactly his specialty - it could affect his ear and cause an infection. Especially if the ball hits him on the ear directly.

Offline Doc Red

  • Chills before posting and wishes others had too
  • RAWK Supporter
  • Legacy Fan
  • ******
  • Posts: 4,876
  • The eye cannot see what the mind does not know.
Re: 'It's beyond the model': have Liverpool exposed the limits of xG?
« Reply #101 on: August 14, 2020, 06:00:13 pm »
This is it really. Expected goals in isolation can easily be misconstrued and doesn’t paint the whole picture. A week or two ago Dan Kennett posted a thread comparing our goal difference to City’s depending on whether we were winning, drawing, or losing. It’s not a surprise that we were equal or better than City when we were drawing or losing and they blew us away when they were winning. It’s obvious from watching our games that once we take the lead it is all about control. We rarely, if ever, run up the score on our opponents. Here’s the thread for those interested:

https://twitter.com/dankennett/status/1288450451671781376?s=21

That's a really great analysis.
The issue isn't  the numbers or the excess use of stats, it's  that the use, or efficient use, of stats to analyse information still has a long way to go. As the OP highlighted, there are things we do that just isn't measured. We aim to win games, City under Pep want to blow teams away. They can aim for that target because their squad is so deep, on the otherhand we have to balance keeping our core group healthy as efficiently as possibly: maximising their efficiency and minimising  effort when possible.

10 years from how, we'll look at the stats used today and just chortle at the limited analysis we currently herald as high end quality.
The child who is not embraced by the village will burn it down to feel its warmth.
There go my people. I must follow them, for I am their leader.

Offline royhendo

  • RAWK Staff
  • Legacy Fan
  • ******
  • Posts: 253,074
  • We all Live in a Red and White Kop
Re: 'It's beyond the model': have Liverpool exposed the limits of xG?
« Reply #102 on: August 14, 2020, 06:01:13 pm »
Roy, I defined things that you claim help to make it an exact science as examples as to why it isn't, we're just looking at that differently.

It’s a case of the flip side of the same coin, yes. But some people are born analysts, and some people aren’t. It’s that simple. You can’t tell the live owner of an analytical brain not to over analyse any more than you can tell the live owner of a pair of lungs not to breathe.

If the possibilities are infinite, any attempt to distil it down can only ever scratch the surface.

Now your argument will be about nice guys finishing last and how if you want to win you have to look beyond the spectacle and get to the nitty gritty. Noted, and I don't disagree.

You should be a script writer. Where are you getting that from?

...for all I'll be mocked and be told my backward thinking doesn't get it, I'm fine with that too.

Again. What?

On the flipside, the data you trust has probably called quite a few things right too. Which lends me to think that the truth is probably somewhere in the middle as it often is.

There we go. I’m happy with that. Can we talk about the topic now? Namely whether the model has been refined beyond what shows up in the general chat on the subject/misuse/misunderstanding of the use of the metric?

I think we all get you - Hinesy used to make the same point - he wants to just enjoy the joyous and the magical aspects of it. I think some of us just like to analyse - can we do that please?
"Word of the day is 'philodox' (17th century): one who is in love with their own opinion, and who consequently believes that everyone else should share it."  @susie_dent on twitter - https://twitter.com/susie_dent/status/1419683653844668422

Offline Samie

  • The next Pharaoh of Egypt. The Ev of drafting! Rumoured to be the 7th, we may need that old magic back! The Timekeeper, ask him what time the action starts.
  • RAWK Supporter
  • Legacy Fan
  • ******
  • Posts: 66,424
Re: 'It's beyond the model': have Liverpool exposed the limits of xG?
« Reply #103 on: August 14, 2020, 06:07:57 pm »
Yes, please answer Lobo's question above Roymundo.

Offline FlashGordon

  • RAWK Supporter
  • Legacy Fan
  • ******
  • Posts: 12,725
  • RAWK Cheltenham 2021 Champion Tipster*
Re: 'It's beyond the model': have Liverpool exposed the limits of xG?
« Reply #104 on: August 14, 2020, 06:09:52 pm »
If a Confident Mane leaves Liverpool at 3PM travelling at 60mph and an Unconfident Mane leaves Glasgow at 5PM travelling at 90MPH, does the Confident Mane make a sound?

Does it even matter if there's no one around to hear it?
So bloody what? If you watch football to be absolutely miserable then go watch cricket.

Offline afc tukrish

  • How long for them sausages? Maggie May's Mythical Turkish Delight. RAWK's Expert Sausage Monster! Oakley Cannonier is fucking boss. Likes blowing his friends and undoing their nuts? Who nose?!
  • RAWK Supporter
  • Legacy Fan
  • ******
  • Posts: 16,933
  • This looks like a nice spot...
    • Flat Back Four
Re: 'It's beyond the model': have Liverpool exposed the limits of xG?
« Reply #105 on: August 14, 2020, 06:14:22 pm »
Does it even matter if there's no one around to hear it?

"I'm only going by what I can see hear with my own two eyes ears..." :D
Since haste quite Schorsch, but Liverpool are genuine fight pigs...

Offline Fiasco

  • Just add water to foam at the mouth. Can't spell San Francisco. Has promised to eat his own cock. Cannibal Self-Harm in that case.....
  • RAWK Supporter
  • Legacy Fan
  • ******
  • Posts: 22,257
  • JFT96.
Re: 'It's beyond the model': have Liverpool exposed the limits of xG?
« Reply #106 on: August 14, 2020, 06:42:46 pm »
It’s a case of the flip side of the same coin, yes. But some people are born analysts, and some people aren’t. It’s that simple. You can’t tell the live owner of an analytical brain not to over analyse any more than you can tell the live owner of a pair of lungs not to breathe.

If the possibilities are infinite, any attempt to distil it down can only ever scratch the surface.

You should be a script writer. Where are you getting that from?

Again. What?

There we go. I’m happy with that. Can we talk about the topic now? Namely whether the model has been refined beyond what shows up in the general chat on the subject/misuse/misunderstanding of the use of the metric?

I think we all get you - Hinesy used to make the same point - he wants to just enjoy the joyous and the magical aspects of it. I think some of us just like to analyse - can we do that please?

Fair enough mate, I didn't mean to come across slightly abrupt in those exchanges, I do find it all interesting regardless. And as I've said numbers and the like isn't my strong point and I do try and read and get a better grasp on it, even if my basic position is what it is.

I suppose having seen Istanbul and been in Cardiff in 2006 and having witnessed our countless comebacks and the like I just try and enjoy it for what it is. Because literally anything can happen in sport. Anyway, I'll let you fucking analysts do your thing :D

Offline newterp

  • RAWK Supporter
  • Legacy Fan
  • ******
  • Posts: 26,769
Re: 'It's beyond the model': have Liverpool exposed the limits of xG?
« Reply #107 on: August 14, 2020, 06:43:07 pm »
If a Confident Mane leaves Liverpool at 3PM travelling at 60mph and an Unconfident Mane leaves Glasgow at 5PM travelling at 90MPH, does the Confident Mane make a sound?

It all depends on if the bears are still shitting in the woods!

Offline El Lobo

  • Chief Suck Up. Feel his breath on your face. Toxic, pathetic, arse-faced, weaselling slimeball. RAWK Maths Genius 2022.
  • RAWK Supporter
  • Legacy Fan
  • ******
  • Posts: 54,990
  • Pretty, pretty, pretty pretty good
Re: 'It's beyond the model': have Liverpool exposed the limits of xG?
« Reply #108 on: August 14, 2020, 06:45:11 pm »
It all depends on if the bears are still shitting in the woods!

I know XB (expected bears) is something Roy is a big champion of
If he's being asked to head the ball too frequently - which isn't exactly his specialty - it could affect his ear and cause an infection. Especially if the ball hits him on the ear directly.

Offline royhendo

  • RAWK Staff
  • Legacy Fan
  • ******
  • Posts: 253,074
  • We all Live in a Red and White Kop
Re: 'It's beyond the model': have Liverpool exposed the limits of xG?
« Reply #109 on: August 14, 2020, 07:18:09 pm »
Fair enough mate, I didn't mean to come across slightly abrupt in those exchanges, I do find it all interesting regardless. And as I've said numbers and the like isn't my strong point and I do try and read and get a better grasp on it, even if my basic position is what it is.

I suppose having seen Istanbul and been in Cardiff in 2006 and having witnessed our countless comebacks and the like I just try and enjoy it for what it is. Because literally anything can happen in sport. Anyway, I'll let you fucking analysts do your thing :D

You come across as sound mate, it’s all good. :)

And the forum does need more joy and magic in it - that’s what makes the club tick more than anything else, you’re right.
"Word of the day is 'philodox' (17th century): one who is in love with their own opinion, and who consequently believes that everyone else should share it."  @susie_dent on twitter - https://twitter.com/susie_dent/status/1419683653844668422

Offline Dave McCoy

  • "Don't you know who I am?!"
  • Legacy Fan
  • ******
  • Posts: 7,461
  • 11,053ft up
Re: 'It's beyond the model': have Liverpool exposed the limits of xG?
« Reply #110 on: August 14, 2020, 07:42:31 pm »
People intuitively talk about xG when they say "so and so should have scored" or that was a "worldie" in that they realize naturally some things are rare and it was amazing to see and others where anybody would score a chance with an open goal from 2 yards out.  It's just a way to quantify and either confirm or deny your intuition at a base level.  What you do with it from there is a whole different discussion.  But it has certainly already changed the way the game is played in that you see certain types of shots barely happening or crosses from certain areas usually forgone due to how the data has shown their usefulness or lack thereof in a teams scoring chances.
« Last Edit: August 14, 2020, 07:44:13 pm by Dave McCoy »

Offline royhendo

  • RAWK Staff
  • Legacy Fan
  • ******
  • Posts: 253,074
  • We all Live in a Red and White Kop
Re: 'It's beyond the model': have Liverpool exposed the limits of xG?
« Reply #111 on: August 17, 2020, 09:25:56 am »
Yeah - Balague talking about the what ifs after Sterling missed that open goal. :D
"Word of the day is 'philodox' (17th century): one who is in love with their own opinion, and who consequently believes that everyone else should share it."  @susie_dent on twitter - https://twitter.com/susie_dent/status/1419683653844668422

Offline Bjornar

  • Kopite
  • *****
  • Posts: 903
Re: 'It's beyond the model': have Liverpool exposed the limits of xG?
« Reply #112 on: August 17, 2020, 06:26:10 pm »
Great thread, with lots of good points from both/all sides of the debate.
 
I don't find it makes sense as a follower of football/LFC to be "for" or "against" analytics in general (it's clearly here to stay anyway, and we use it to our advantage currently), for me it's more a case of where the line is between sensible versus misleading reading of the particular stats in a particular context. RAWK is great for learning more about that, but personally I do get annoyed sometimes when xG or player radars are used in LFC discussions to support arguments that stray too much into the territory of "Klopp needs to make this specific change, the stats clearly show it", for my liking at least.   

Offline Dave McCoy

  • "Don't you know who I am?!"
  • Legacy Fan
  • ******
  • Posts: 7,461
  • 11,053ft up
Re: 'It's beyond the model': have Liverpool exposed the limits of xG?
« Reply #113 on: August 17, 2020, 10:02:55 pm »
Great thread, with lots of good points from both/all sides of the debate.
 
I don't find it makes sense as a follower of football/LFC to be "for" or "against" analytics in general (it's clearly here to stay anyway, and we use it to our advantage currently), for me it's more a case of where the line is between sensible versus misleading reading of the particular stats in a particular context. RAWK is great for learning more about that, but personally I do get annoyed sometimes when xG or player radars are used in LFC discussions to support arguments that stray too much into the territory of "Klopp needs to make this specific change, the stats clearly show it", for my liking at least.

People have been making those types arguments for ages regardless of "stats".  Just with stats you actually may be right once in a while instead of wrong all the time.

Offline Redwhiteandnotblue

  • God's spin doctor.
  • Legacy Fan
  • ******
  • Posts: 3,017
Re: 'It's beyond the model': have Liverpool exposed the limits of xG?
« Reply #114 on: August 19, 2020, 07:39:09 pm »
It all depends on if the bears are still shitting in the woods!

I thought it was the Pope? Is the bear a Catholic?

Offline afc tukrish

  • How long for them sausages? Maggie May's Mythical Turkish Delight. RAWK's Expert Sausage Monster! Oakley Cannonier is fucking boss. Likes blowing his friends and undoing their nuts? Who nose?!
  • RAWK Supporter
  • Legacy Fan
  • ******
  • Posts: 16,933
  • This looks like a nice spot...
    • Flat Back Four
Re: 'It's beyond the model': have Liverpool exposed the limits of xG?
« Reply #115 on: August 19, 2020, 08:13:29 pm »
I thought it was the Pope? Is the bear a Catholic?

the Transfer Forum is Catholic?  :o
Since haste quite Schorsch, but Liverpool are genuine fight pigs...

Offline KissThisGuy

  • OnlyThroughAFaceMaskThough
  • Legacy Fan
  • ******
  • Posts: 1,921
  • It was raining, I went outside, I got wet.
Re: 'It's beyond the model': have Liverpool exposed the limits of xG?
« Reply #116 on: August 20, 2020, 03:28:51 pm »
People intuitively talk about xG when they say "so and so should have scored" or that was a "worldie" in that they realize naturally some things are rare and it was amazing to see and others where anybody would score a chance with an open goal from 2 yards out.  It's just a way to quantify and either confirm or deny your intuition at a base level.  What you do with it from there is a whole different discussion.  But it has certainly already changed the way the game is played in that you see certain types of shots barely happening or crosses from certain areas usually forgone due to how the data has shown their usefulness or lack thereof in a teams scoring chances.
Sterling dislikes

Offline Rhi

  • Rhisuscitated
  • RAWK Staff
  • Legacy Fan
  • ******
  • Posts: 1,952
Re: 'It's beyond the model': have Liverpool exposed the limits of xG?
« Reply #117 on: August 21, 2020, 04:05:50 pm »
Why Liverpool hired a French start-up to turn video into data

Quote
PARIS start-up SkillCorner had just finished working for betting clients at the 2018 World Cup when the call came from Liverpool’s Head of Research, Ian Graham.

“He’d seen some of our data visualisations and said: 'If you’re doing this from broadcast for the World Cup then this could be interesting for us in player recruitment,'"¯ remembers the French company’s co-founder, Hugo Bordigoni.

At that time, SkillCorner was providing betting company websites with live animations showing the location of the ball and players during football matches. Liverpool spent a year helping them develop this into a performance product, before signing a contract a year ago.

So what was it that Graham had first seen in this relatively obscure French company that had previously counted only betting companies among its clients?

The answer, in short, is tracking data, that goldmine of positional co-ordinates that clever data scientists can turn into valuable insights.

Whereas traditional event data tells you what happens on the ball during a game, such as the number of passes, tackles and shots, tracking data reveals the position and movements of every single player in every single moment of a match.

This totals more than a million data points per 90 minutes, enabling a data scientist to calculate things such as the pressure on a player when they made a particular pass, whether they broke the lines, and if they made the best possible decision in a given situation.

This is all crucial information when a club is trying to build a comprehensive picture of the capabilities of a potential transfer target.

The Premier League is about the embark on the second season of a three-season deal with US firm Second Spectrum to provide tracking data for all its games. Data is captured at 25 frames-per-second by a series of cameras dotted around every stadium.

Eleven other European leagues have tracking data deals, with either ChyronHego, STATSPerform, Second Spectrum or InStat.

So far, so good. But the problem for the likes of Liverpool is that these leagues only share their tracking data with their own clubs.

“Liverpool were happy with the tracking data they got for the Premier League and Champions League,” Bordigoni explains, “but if they wanted to scout players in France, Spain, Germany and beyond - which of course they do - then they couldn’t access the data from those countries.

"Access to the data belongs to each particular league and they don’t share it.”
This is where SkillCorner come in. They are different to the likes of Second Spectrum, because they derive tracking data from HD video footage rather than from their own cameras in stadiums.

They are able to do this with any HD video feed, but mainly use video scouting platform Wyscout, with whom they have a partnership. SkillCorner currently gather tracking data from 23 global competitions (21 leagues plus the Champions League and Europa League) and aim to cover as many as 40 within the next 18 months.

Now Liverpool can get raw tracking data (giving the x and y co-ordinates of every player, the ball and referee, plus the z co-ordinates for the ball, at 10 frames per second) for every game in 23 competitions.

This enables them to do detailed analysis on potential transfer targets, compare them, and to benchmark against Jurgen Klopp’s style of play and physical requirements.

After signing up the then European champions as their first football client, SkillCorner now have deals with clubs in England, France, Germany and Italy and are in advanced talks with others.

I've only quoted half of the article there. See the rest here:
https://trainingground.guru/articles/why-liverpool-hired-french-start-up-to-turn-video-into-data

It's this sort of thing that those who chuck stats about out of context don't understand. Just seen some kid on Twitter conclude that Gini Wijnaldum should be let go because his xG is crap and goal involvements is crap and all this sort of absolute rubbish. It's obvious to anyone with half a brain and an interest in the game that Gini is so important to how we play. The club can quantify that using things like positional data and resistance to pressure on the ball and all this kind of stuff (along with stuff that would blow our minds, I'm sure).

I've followed this thread closely over the last week. I sympathise with those who can't be arsed with the whole xG and stats thing, I really do. As a football fan I TOTALLY get why you just want to watch the game in all its wonderful unpredictability. I get why you don't want to see the fun maths-ed out of it. I also usually find myself wanting to throw my laptop out the window when someone uses xG (or whatever other stat to be honest) to say the result of a game was wrong.

Then there's the data scientist in me who would fucking love an hour with Ian Graham, because what our guys are doing is so far ahead of what's available in mainstream football analytics.
“Above all, I would like to be remembered as a man who was selfless, who strove and worried so that others could share the glory, and who built up a family of people who could hold their heads up high and say 'We're Liverpool'.” - Bill Shankly

Offline Bjornar

  • Kopite
  • *****
  • Posts: 903
Re: 'It's beyond the model': have Liverpool exposed the limits of xG?
« Reply #118 on: August 21, 2020, 07:46:27 pm »
People have been making those types arguments for ages regardless of "stats".  Just with stats you actually may be right once in a while instead of wrong all the time.

I disagree, there can be a particular kind of over-confidence and surface impressiveness that come with arguments relying on stats in football discussion IMO, which isn't there in the same way when someone is just offering an opinion. That's talking about football media and fan forums, not the work the actual experts are doing.

Several examples on RAWK for example of arguments (over)-relying on stats that IMO hardly meant being less likely to be wrong, especially before this season:

- crosses were an ineffective way of scoring, so Trent and Robertson wouldn't be able to sustain their production
- City's xG compared to ours meant that our best bet to beat them was moving closer towards their playing style
- we couldn't keep playing with three workmanlike midfielders and keep winning, needed a "midfield progressor" and/or better individual production numbers from midfield
- we had to sign someone to not "crash hard", Understat xG tables proved it and RAWK posters bought that that was thought-provoking and persuasive analysis (see the "bold predictions" thread)

So I think it's valid to be a bit sceptical about supposed "statistical analysis" in football/fan media. But again, not doubting the value of analytics when it's the real thing and not just quoting some of the numbers and terminology. Think we're all still learning how to apply this in a sensible way as we go along, both the enthusiasts and the sceptics.


Great post.
« Last Edit: August 21, 2020, 07:55:08 pm by Bjornar »

Offline Dave McCoy

  • "Don't you know who I am?!"
  • Legacy Fan
  • ******
  • Posts: 7,461
  • 11,053ft up
Re: 'It's beyond the model': have Liverpool exposed the limits of xG?
« Reply #119 on: August 21, 2020, 09:02:38 pm »
I disagree, there can be a particular kind of over-confidence and surface impressiveness that come with arguments relying on stats in football discussion IMO, which isn't there in the same way when someone is just offering an opinion. That's talking about football media and fan forums, not the work the actual experts are doing.

Several examples on RAWK for example of arguments (over)-relying on stats that IMO hardly meant being less likely to be wrong, especially before this season:

- crosses were an ineffective way of scoring, so Trent and Robertson wouldn't be able to sustain their production
- City's xG compared to ours meant that our best bet to beat them was moving closer towards their playing style
- we couldn't keep playing with three workmanlike midfielders and keep winning, needed a "midfield progressor" and/or better individual production numbers from midfield
- we had to sign someone to not "crash hard", Understat xG tables proved it and RAWK posters bought that that was thought-provoking and persuasive analysis (see the "bold predictions" thread)

So I think it's valid to be a bit sceptical about supposed "statistical analysis" in football/fan media. But again, not doubting the value of analytics when it's the real thing and not just quoting some of the numbers and terminology. Think we're all still learning how to apply this in a sensible way as we go along, both the enthusiasts and the sceptics.

Great post.

Over-confidence that the stats are correct is fine, they're actual facts.  Over-confidence in what the stats mean and you're issue with that is basically two sides of the same coin.  You can't tell me you don't know people that aren't just as forceful in their opinions regardless of stats, that's just human nature.  As far as your examples:

Crossing - What make Robbo and Trent unique in their crossing that they buck statistical facts?  How likely is that to continue?
ManC's xG - Regardless of style wouldn't you say it's sub-optimal to not be consistently creating the same number of chances?
Midfield - this sounds like nothing to do with facts/stats
"crash hard" - again, opinions?

It's valid to be skeptical of peoples opinion about what fact/stats mean but your examples are kind of that and kind of not.