*** OMT: Tottenham Hotspur vs Leicester City ***

[nayimfromthehalfwayline]

Classic - citing an opinion, while explicitly stating that you've chosen not to investigate whether or not it's valid. (Not classic of you, just people in general )

At this point. Finding how xg is measured isnt of any particular interest. And seeing that it doesnt actually reflect the game (as it was presented) supports the notion it isnt as fool proof as perhaps believed.

I might look into it at a later point, depends how much it gets used in debate. If people try using it as definitive proof of anything itll be good to know its ins and outs more thoroughly.

Im pretty sure youve been around the site long enough to have seen for yourself there are a number of posters that see stats as ultimate proof of their pov. Not realising that stats can be used to prove basically any pov.

I see them as a tool, no more. Its interesting to think, for example, a player was operating in a different way than usual, and then seeing a heatmap that supports/contests that idea.

Over all, for me, context is everything with stats. As the Arsenal game example above was trying to show.

 

[FatBloke]

Well take away their pen that shouldn't have been, and add our pen that should, and what does that do for your xG?

Absolutely nothing, as xG doesn't (and cannot) take penalties into account. At least, Caley's xG doesn't -- there seems to be a growing number of alternative xG methods but I think Caley's was the first. He usually puts penalties (at least, those that are scored) in brackets after his xG numbers.

Caley's xG gives a figure for each chance created between 0 (never going to be a goal, ever) and 1 (certain goal), based on where the chance is (middle of the 6-yard box vs 30 yards out), what sort of chance it is (e.g. header vs over-the-shoulder volley) and how it came about (e.g. through-ball vs corner kick). The figures come from years of stored analysis -- there must be millions of data points. Add those numbers and you get something that should approximate a game's scoreline.

The major failing of xG is that it doesn't take into account who's taking (or wasting) each chance. So the disparity between xG and the actual score is an indicator of the quality of a side's attack -- as Parklane (I think) commented, you'd expect Sonny and Eriksen to take their chances better than Barnes. A team or player on a hot streak (think Leicester in their Prem-winning season or Salah in 2017/18) will exceed their xG, while a team that couldn't score in a brothel will do worse (Huddersfield, I'm looking at you). Another exception is where a team has a fantastic shot-stopper -- ManU often concede far fewer than xG suggests they would, for example.

I think of xG as similar to the possession stat, esp. possession separated into thirds of the pitch. It's not a sure-fire indicator (nothing is except the final score!), but it's an interesting tool to help analyse what happened in a given match.

What xG does in Sunday's match confirms my feeling as I was watching the game on TV -- Leicester were making plenty of decent-quality chances and would have won if they were more clinical. That's a worry and we wouldn't expect a top 6 team to be so profligate.

 

[elltrev]

At this point. Finding how xg is measured isnt of any particular interest. And seeing that it doesnt actually reflect the game (as it was presented) supports the notion it isnt as fool proof as perhaps believed.

I might look into it at a later point, depends how much it gets used in debate. If people try using it as definitive proof of anything itll be good to know its ins and outs more thoroughly.

Im pretty sure youve been around the site long enough to have seen for yourself there are a number of posters that see stats as ultimate proof of their pov. Not realising that stats can be used to prove basically any pov.

I see them as a tool, no more. Its interesting to think, for example, a player was operating in a different way than usual, and then seeing a heatmap that supports/contests that idea.

Over all, for me, context is everything with stats. As the Arsenal game example above was trying to show.

Fair play for a calm and balanced response to my slightly inflammatory message

Ironically I think the majority of people are probably on the same page about stats - that they're a tool that can be useful to understand reality and support arguments, but virtually never represent objective and all-encompassing truth, and should always be interrogated carefully. To be honest I take issue with either extreme outside of that - I just tend to find that it's the people who are more 'anti-stats' who are a) more commonly found, and b) more extreme in their assessment of the value of stats (or lack thereof).

 

[FatBloke]

Here's an example, come to think of it. Team A decides to play deep, letting Team B have possession but countering it with a well-organised defence and 8 or 9 men behind the ball. Team A's plan is to hit them on the counter with their pacey forwards.

In such a game it wouldn't be unusual for Team B to have 70%+ possession and 20+ shots, as opposed to Team A only ending up with 5 or 6. So all the conventional stats would favour Team B.

xG would show that Team B's shots, mostly from distance, are less likely to score than Team A's shots, where 3 or 4 attackers have steamed down the pitch or someone's played a quality ball over the top.

In that scenario, I think xG would paint a much more realistic picture of the match than the conventional stats would.

 

[harr1984]

Ignoring the xG stats watching the game Leicester had the better chances, I dont see how that could be argued. Whilst Barnes isnt anywhere yet near the level of a Son/Eriksen he is playing as an AM, and they quite clearly carved us open on a good few occassions. Our goals were from a set piece, a great strike from Eriksen outside the area and a counter attack when Leicester were pushing forward for the equaliser - we werent particularly creative throughout the game at all in comparison to what we can be. We were second best in terms of overall play, but were clearly more clinical and if you dont finish your chances you are going to lose against the top sides. I dont understand why Leicester fans dont seem keen on Puel, he has them playing some decent football and has gathered a few promising young footballers there....

 

[elltrev]

Here's an example, come to think of it. Team A decides to play deep, letting Team B have possession but countering it with a well-organised defence and 8 or 9 men behind the ball. Team A's plan is to hit them on the counter with their pacey forwards.

In such a game it wouldn't be unusual for Team B to have 70%+ possession and 20+ shots, as opposed to Team A only ending up with 5 or 6. So all the conventional stats would favour Team B.

xG would show that Team B's shots, mostly from distance, are less likely to score than Team A's shots, where 3 or 4 attackers have steamed down the pitch or someone's played a quality ball over the top.

In that scenario, I think xG would paint a much more realistic picture of the match than the conventional stats would.

This is what I find interesting with XG - it attracts some negativity, despite it surely being a more interesting and useful stat than possession, territory, shots, shots on target, corners etc (all of which you can also see with your own eyes) - how do people feel about those stats?!

Finally, if XG stops pundits such as Shearer constantly using the cliche 'they should have had 5 or 6 today', then it will have all been worthwhile for me.

 

[nayimfromthehalfwayline]

Fair play for a calm and balanced response to my slightly inflammatory message

Ironically I think the majority of people are probably on the same page about stats - that they're a tool that can be useful to understand reality and support arguments, but virtually never represent objective and all-encompassing truth, and should always be interrogated carefully. To be honest I take issue with either extreme outside of that - I just tend to find that it's the people who are more 'anti-stats' who are a) more commonly found, and b) more extreme in their assessment of the value of stats (or lack thereof).

Im certainly not one of those. I tend to only become a bit 'anti stats' in the face of someone using them as an absolute, which they very rarely are.

Otherwise Im generally like you and somewhere in the middle ground.

I never thought a game with Leicester would end up producing a conversation like this, its funny. Sure, Leicester made some good chances but I at no point felt we were losing the game/initiative or that they were a real danger to us.

Yes it was pretty end to end, but I felt we were resolute.
Yes we didnt create chance after chance, but I felt we carried more threat when we broke.
Yes, lots of pretty passing and runs from them, but I felt they rarely amounted to anything...

Yes, I can see they made some good chances, but even so...

And yet, look how other people view it.

 

[alekaras]

Well take away their pen that shouldn't have been, and add our pen that should, and what does that do for your xG?

He already has taken away their pen.

 

[alekaras]

I think @Raziel nails it, to be honest.

And @parklane1 . I saw the game, and at no point felt "we've got away with one here". Which I should had we been so obviously outplayed.

That looks like 4 nailed on chances right in front of goal. Id guess Grays hilarious header would be one of them, which stacks up to Raziels point. Another I remember shot right at Lloris, made it easy on him. Theres another.

I dont actually know how xg decides upon the quality of a chance, but I suspect it is completely lacking in context.

I often refer back to the first time we beat Arsenal at the Lane under Redknapp. We deliberately sat very deep and countered. Defended across the box and pushed their play wide where they were useless. And beat them, actually quite comfortably, thanks to a very astute game plan.

The stats would have it we were utterly outclassed in every respect. They will have had going on 80% possession, shots into the double figures, 2x (or3) the passes we had, most of the play in our third...

And yet the reality is we were comfortable, in control, and frankly the better team on the day fully deserving of the three points.

Just because you don't understand it doesn't mean you have to poo poo it. xG looks at thousands of similar chances and assigns a value to them. Shots from outside the box have a low value. So in the Arse game you mentioned, I bet the xG was in our favor, if that was all they were doing. Leicester, however, carved us apart several times and had much higher quality chances. We were lucky they didn't convert them and it's as simple as that.

 

[nayimfromthehalfwayline]

Just because you don't understand it doesn't mean you have to poo poo it. xG looks at thousands of similar chances and assigns a value to them. Shots from outside the box have a low value. So in the Arse game you mentioned, I bet the xG was in our favor, if that was all they were doing. Leicester, however, carved us apart several times and had much higher quality chances. We were lucky they didn't convert them and it's as simple as that.

Bit touchy on the subject are we?

I do not feel Leicester "played us off the park", as was the original comment, and looking at a pretty xg picture doesnt change that.

Yes, some good chances, I can see that. I feel, as was pointed out by others, who the chance fell to is in fact relevant - and unaccounted for. And I also feel, regardless, some good chances doesnt equate to "played us off the park".

And after all that, I havent actually poo poo'd it, have I?

 

[StephenH]

Finally, if XG stops pundits such as Shearer constantly using the cliche 'they should have had 5 or 6 today', then it will have all been worthwhile for me.

I like Shearer having chunter on motd.
It will be bloody boring quoting stats ad nauseam, I’ll stick to the Geordie Bard!

 

[FatBloke]

I think @Raziel nails it, to be honest.

And @parklane1 . I saw the game, and at no point felt "we've got away with one here". Which I should had we been so obviously outplayed.

That looks like 4 nailed on chances right in front of goal. Id guess Grays hilarious header would be one of them, which stacks up to Raziels point. Another I remember shot right at Lloris, made it easy on him. Theres another.

I dont actually know how xg decides upon the quality of a chance, but I suspect it is completely lacking in context.

I often refer back to the first time we beat Arsenal at the Lane under Redknapp. We deliberately sat very deep and countered. Defended across the box and pushed their play wide where they were useless. And beat them, actually quite comfortably, thanks to a very astute game plan.

The stats would have it we were utterly outclassed in every respect. They will have had going on 80% possession, shots into the double figures, 2x (or3) the passes we had, most of the play in our third...

And yet the reality is we were comfortable, in control, and frankly the better team on the day fully deserving of the three points.

That was the DANNY ROSE ON DEBUT game in 2010/11, wasn't it? Scary that it's almost a decade ago now.

Anyway, Alekeras quoting your post highlighted that that game shows almost exactly the example scenario I gave a few posts back. Arsenal hammered us in terms of possession and shots at goal, but most of their shots will have been low quality -- far away from goal, hopeful balls pinged in from wide positions, and so on. xG didn't exist back then, but I'm confident that it would have shown a much closer game than the conventional stats did. I doubt it would show us on top though as, let's face it, young Danny's thunderous strike was a very low-probability chance.

As you say, it's been an interesting discussion.

Michael Caley (who's a Spurs fan in the US, BTW) gives an in-depth discussion of his xG calculations here:
https://cartilagefreecaptain.sbnati...ier-league-projections-and-new-expected-goals

I suspect that he's refined it further in the last couple of years, but that's the basics.

 

[nayimfromthehalfwayline]

Yeah I think that was the one.

Im aware of what xg is, and what a positive one means, and also that it broadly correlates to the actual scores.

What I refute is that it shows a team being "played off the park", and also question its accuracy* given it takes no account of who the chances fell to.

Leicester had 4 good chances. They may well have had their asses handed to them in all other aspects of the game, and lost 3-1 - most would say deservedly. Would a better xg on their part offer proof they "played off the park"? I venture not.

Which plays into the wider conversation about stats. In so much as, they can only really support a point of view and not much else. Also, Im a big believer that a single stat in isolation proves nothing really, it lacks context.



*accepting its calculation and assumption and open to error. And looking at his explanation, it seems much of it is still open to his interpretaion of extrapolating opta data.

 

[parklane1]

When I first read that, I thought you were talking about your virginity!

 

[Craig_J]

Too much?....

View attachment 5971

You should bump that match thread every now and then exclaiming how hurt you still are by the defeat, regardless as to Leicester's forms and results. The crack in your heart that shall forever remain.

 

[alekaras]

Bit touchy on the subject are we?

I do not feel Leicester "played us off the park", as was the original comment, and looking at a pretty xg picture doesnt change that.

Yes, some good chances, I can see that. I feel, as was pointed out by others, who the chance fell to is in fact relevant - and unaccounted for. And I also feel, regardless, some good chances doesnt equate to "played us off the park".

And after all that, I havent actually poo poo'd it, have I?

About as touchy as when my dad insists I take a certain route, because he always goes that way and it's fast, when Google Maps tells me it's bumper to bumper. The game I saw had my heart in my mouth every time Leicester created a gilt-edged chance and breathing a sigh of relief when they missed and the xG pretty much confirmed that. And then there are other games I felt we were under the kosh only to look at the xG and realize I was worrying about chances the opposition created that had very little probability of being converted.

Now if your brain works in such a way that you can assign a quality number to each chance, tally it up, and keep a running total while the game is going on, then kudos to you. For the rest of us normal people it's helpful to look at other pieces of data.

 

[alekaras]

You should bump that match thread every now and then exclaiming how hurt you still are by the defeat, regardless as to Leicester's forms and results. The crack in your heart that shall forever remain.

Brilliant idea! I haven't laughed so hard after reading @Jon 's Foxes Talk trolling in a long time!

 

[braineclipse]

Yeah I think that was the one.

Im aware of what xg is, and what a positive one means, and also that it broadly correlates to the actual scores.

What I refute is that it shows a team being "played off the park", and also question its accuracy* given it takes no account of who the chances fell to.

Leicester had 4 good chances. They may well have had their asses handed to them in all other aspects of the game, and lost 3-1 - most would say deservedly. Would a better xg on their part offer proof they "played off the park"? I venture not.

Which plays into the wider conversation about stats. In so much as, they can only really support a point of view and not much else. Also, Im a big believer that a single stat in isolation proves nothing really, it lacks context.



*accepting its calculation and assumption and open to error. And looking at his explanation, it seems much of it is still open to his interpretaion of extrapolating opta data.

Most pundits I have heard agree that Leicester deserved to get something from that game.

Xg has it's weaknesses for sure, but so does the evaluation everyone has from just watching the game. In this case I agree with the stat that we were a bit lucky. But we've been unlucky in other games.

Hopefully our performances will improve quickly. I think we've played quite well recently, but the Leicester game was our weakest performance for a while now.

 

[nayimfromthehalfwayline]

About as touchy as when my dad insists I take a certain route, because he always goes that way and it's fast, when Google Maps tells me it's bumper to bumper. The game I saw had my heart in my mouth every time Leicester created a gilt-edged chance and breathing a sigh of relief when they missed and the xG pretty much confirmed that. And then there are other games I felt we were under the kosh only to look at the xG and realize I was worrying about chances the opposition created that had very little probability of being converted.

Now if your brain works in such a way that you can assign a quality number to each chance, tally it up, and keep a running total while the game is going on, then kudos to you. For the rest of us normal people it's helpful to look at other pieces of data.

Let me be absolutely clear. I do not consider myself some sort of genius, I didnt sit watching the game assuming I was seeing something others werent.

I just watched it. And I didnt feel under threat at any point really (and I often do). They enjoyed a lot of possession and got a lot of positive comment from the commentary (particularly Tielemans) but to me most of it was just unproductive.

And all the while Gray was up front they couldnt score in a brothel.

That was it.

Then people tell me we were played off the park, and a diagram showing 4 good chances (positions maybe more appropriate?) as evidence of such - and I just dont buy it.

 

[glorygloryeze]

What an important win; REALLY puts us right in the mix, especially if we can follow it up with a win away at Burnley in our next league match. Imagine if that was to happen AND Kane was to return AND Liverpool hit the skids vs Man Utd??

Ok i'll stop now...

Anyway, COYFS