The xG nonsense

[SuperKevsLeftBoot]

xG is simply the chance of the goal being converted by an AVERAGE player. The value doesn't tell us anything else.

That’s fair enough but I would like to think our players are better than average.

Kane has outperformed his xg every season bar 1 and Jesus has never hit it.

So basically it’s telling us that our players are far better than the average as we’re outperforming it while other teams like Leeds aren’t. Which i agree with.

Yes

Aye for a man average striker

 

[johnkaysleftleg]

Statistically if we have the same relative xG next season then we will almost certainly be in the bottom 3 or 4.

If you look a the way we progressed that is unlikely, first half of the season or Xg for was 0.78 per game, quite worrying and unsustainable to a team wanting to stay up long term, the second half it was 1.25 per game which is perfectly fine. Defensively we were consistent at about 1.39 per game.
Anyone who has watched us can see how we have progressed in terms of creating chances as the season has progressed while largely holding it together at the back.

 

[Kent Boy 2]

So it’s based on players ability as well then? Coz if Wilson had all 3 he’d have have scored two if not all 3.

If it takes into account the player then I’d understand it more. Brobbeys finishing isn’t the best.. Wilson is far better

You mean like he scored the 1v1 in the last minute at Liverpool?

xG uses all sorts of previous data and takes into account multiple of the immediate circumstances to create a model. That model is pretty accurate. Under the circumstances that it occurred, and using previous data from across the PL over the past couple of seasons, I don’t think that Brobbey chance would be taken more than 1 time in 3 - not even by Haaland.

If you look a the way we progressed that is unlikely, first half of the season or Xg for was 0.78 per game, quite worrying and unsustainable to a team wanting to stay up long term, the second half it was 1.25 per game which is perfectly fine. Defensively we were consistent at about 1.39 per game.
Anyone who has watched us can see how we have progressed in terms of creating chances as the season has progressed while largely holding it together at the back.

What do you mean by ‘unlikely’?

If you mean unlikely that we will finish next season with an xG as low as this season, then I 100% agree. We have been improving all season in our forward play, and have data analytics and recruitment teams who will be using data such as this to find areas of improvement and players who can help make those improvements.

If you mean unlikely that we will finish in the bottom 3-4 if we finish next season with the same season long average xG as we finished this season then I 100% disagree. We were a massive statistical anomaly this season, there almost certainly won’t be the same anomaly next season.

 

[Charmless Man]

If you look a the way we progressed that is unlikely, first half of the season or Xg for was 0.78 per game, quite worrying and unsustainable to a team wanting to stay up long term, the second half it was 1.25 per game which is perfectly fine. Defensively we were consistent at about 1.39 per game.
Anyone who has watched us can see how we have progressed in terms of creating chances as the season has progressed while largely holding it together at the back.

And seeing a progression in the trend is good, useful data.

 

[DJ Doubt]

But thats only half the story. We didnt conceded as much either. Only 3 other teams let in fewer goals than us

Which all falls into line with a team finishing 7th but being bottom of the XG table

Why is this so hard to grasp?

If Sunderland progress over the next few seasons, I'd expect to see us rise up the XG table as our attack gets stronger. It will be an indication of attacking progress if we do.

 

[johnkaysleftleg]

You mean like he scored the 1v1 in the last minute at Liverpool?

xG uses all sorts of previous data and takes into account multiple of the immediate circumstances to create a model. That model is pretty accurate. Under the circumstances that it occurred, and using previous data from across the PL over the past couple of seasons, I don’t think that Brobbey chance would be taken more than 1 time in 3 - not even by Haaland.

What do you mean by ‘unlikely’?

If you mean unlikely that we will finish next season with an xG as low as this season, then I 100% agree. We have been improving all season in our forward play, and have data analytics and recruitment teams who will be using data such as this to find areas of improvement and players who can help make those improvements.

If you mean unlikely that we will finish in the bottom 3-4 if we finish next season with the same season long average xG as we finished this season then I 100% disagree. We were a massive statistical anomaly this season, there almost certainly won’t be the same anomaly next season.

I'm saying it's unlikely we'll have the same negative Xg offset next season due to the way we have progressed and therefore unlikley well be bottom 3 or 4.

 

[EternallySAFC]

xG is simply the chance of the goal being converted by an AVERAGE player. The value doesn't tell us anything else.

That’s fair enough but I would like to think our players are better than average.

Kane has outperformed his xg every season bar 1 and Jesus has never hit it.

So basically it’s telling us that our players are far better than the average as we’re outperforming it while other teams like Leeds aren’t. Which i agree with.

Aye for a man average striker

Players miss chances. I bet it's rare you come away from a game and think "We shouldn't have scored the number of goals we did" compared to "We should really have scored more than we did". xG is not perfect but it gives a good account for how often a similar chance is scored. Players are also generally playing against others of a similar standard. So a bang average striker against a bang average goalkeeper/defender may score as often as a great striker against a great goalkeeper/defender. If its not for you just ignore it.

 

[Kent Boy 2]

I'm saying it's unlikely we'll have the same negative Xg offset next season due to the way we have progressed and therefore unlikley well be bottom 3 or 4.

Great. Personally I fully agree with you

 

[johnkaysleftleg]

xG is one statistic that should not be used purely on its own, but it is relatively good to help describe a predicted outcome.

I’m not sure why people get so hung up on it being terrible by using one or two examples. Yes, Sunderland were an outlier, but all statistics have multiple outliers - otherwise there would be no betting industry. However, it’s a pretty good indicator when you consider that 5 of the top 6 in the xG table were in the top 6 of the actual table, and the bottom 5 in the actual table were all in the bottom 7 in the xG table.

It’s like saying that scoring more goals that you concede is a bad statistic because Sunderland finished 7th with a GD of -6, whilst Forest finished 16th with a GD of -3

My post above this one shows how it's really not that great to predict outcomes in games, in fact if the Xg is quite close, next to useless.

 

[Kent Boy 2]

My post above this one shows how it's really not that great to predict outcomes in games, in fact if the Xg is quite close, next to useless.

Yet the data shows that >80% of clubs would be in a very similar position in the league at the end of the season if points were awarded for xG difference in each game as they were for actual goal difference in each game. So on a one off basis it may not be great, but over a 380 game data set it is pretty good.

 

[AC Mack]

XG had us down as 13th and Hull as relegated. It's a useless stat for the kind of people who live in their parents spare bedroom.

It has Aston Villa 13th and Sunderland 18th. 9 and 11 places out respectively.

It has Arsenal, Man United, Manchester City and Nottingham Forest in their exact positions. It has Liverpool, Bournemouth, Brentford, Fulham, Everton, West Ham, Wolves and Burnley as 1 place out from where they finished. The last two would have been exact positions too if Burnley hadn't equalised againat Wolves on Sunday.

But you've decided that it's useless based on a couple of outliers.

 

[LineTrash]

The Xg I'm using is the most advanced,

Logon or register to see this image

 

[johnkaysleftleg]

Yet the data shows that >80% of clubs would be in a very similar position in the league at the end of the season if points were awarded for xG difference in each game as they were for actual goal difference in each game. So on a one off basis it may not be great, but over a 380 game data set it is pretty good.

Well in nearly half the games this season (48%) when the XG difference was less that 0.75 it was scarcely better than picking the result out of a hat 36%

 

[Frijj]

There’s a U shaped curve in the misapplication of xG

People at one end who don’t understand it, and people at the other who understand it but misuse it

Those of us in the middle are those who understand it and use properly.

 

[Definitely Maybe]

xG is simply the chance of the goal being converted by an AVERAGE player. The value doesn't tell us anything else.

That’s fair enough but I would like to think our players are better than average.

Kane has outperformed his xg every season bar 1 and Jesus has never hit it.

So basically it’s telling us that our players are far better than the average as we’re outperforming it while other teams like Leeds aren’t. Which i agree with.

Aye for a man average striker

Again though, if we're outperforming our xG it suggests our attacking players are more clinical than average. Which you go on to agree with but then criticise as nonsense?

You keep referencing Kane who is arguably the best striker in the world. Is he the most clinical striker in the world? The answer is no, his goal conversion rate is 22% this season which puts him 35th out of all players in the top 5 European leagues. Will Osula is 3rd at 30%. Kane's conversion rate on previous seasons usually ranges from 17% to 20%, this was the best season of his career.

Is Osula a better goalscorer than Kane? Absolutely not, but what sets them apart isn't the fact Kane is a better finisher, its because Kane's movement and intelligence is on another planet and he consistently gets into great positions where he can get shots away from dangerous areas. There's loads of research which shows the best strikers aren't the ones with the highest conversion rates and the ideas of the best strikers being the best finishers is a bit of a myth, the reality is the best strikers are the ones with the highest xG because they get into better positions than anyone else.

If you think of Sunderland as an example, we've had loads of shit strikers but the only one I can remember who actually missed loads of chances was Altidore. The likes of Grigg and Graham weren't shit because they kept missing chances, they were shit because they were miles off the pace and never got a sniff of anything near the goal.

You also keep mentioning Leeds underperforming xG - who is their main striker? Calvert-Lewin, who historically has always underperformed xG and has had spells this season and throughout his career where everything he touched went in followed by long goal droughts. Leeds and DCL's xG data suggests they're a fairly creative team who aren't great at putting away those chances. Ours suggests we aren't particularly great at creating chances but good at finishing what we do create. There's no positive or negative to either of those beyond that, just an observation. Statistically, teams generally do revert to the mean over the long term but anyone telling you a team should have finished here or there because of xG is talking shit and missed the point of it.

There's a couple of examples that are frequently quoted - in Klopp's final season at Dortmund they had a terrible first half of the season and came in for a lot of criticism, but Klopp pointed out their xG data was almost identical to the season they won the league, they'd just been unlucky and their luck would turn around. He was laughed at but sure enough they had a great second half of the season without doing anything differently. And in our League One season under Jack Ross there was holy hell on on here when pundits suggested we weren't as good as results suggested because we'd been lucky with the likes of McGeady and Maguire scoring screamers and McLaughlin saving everything despite us giving away loads of chances. That George Elek received a load of abuse for predicting Barnsley to go to ahead of us, especially after we beat them 4-2. Sure enough though the screamers stopped going in, we started to concede more and slipped away. Both are examples of teams under or over performing xG and eventually reverting to the mean.

 

[Loch Ness Drogba fan]

I bet forest fans said exactly the same at the end of their last season.

 

[Kent Boy 2]

There’s a U shaped curve in the misapplication of xG

People at one end who don’t understand it, and people at the other who understand it but misuse it

Those of us in the middle are those who understand it and use properly.

Perfect description. Fortunately the majority of football clubs (including ours) have people who are in the middle.

Well in nearly half the games this season (48%) when the XG difference was less that 0.75 it was scarcely better than picking the result out of a hat 36%

What about in the other 52% of games?

 

[TartanMackem]

xG isn’t nonsense but the way people use it is. The very concept of an xG table is utter bollocks.

 

[No7]

I am absolutely delighted football is played on grass and not on a excel sheet

 

[Pacific]

i don't think it's as nonsense as people make it out to be mind

it's a decent indicator and based on how many games this season we've had where it felt like we were creating absolutely nothing i don't think it's that outlandish

certainly wouldn't want to be going into next season with such a defensive mindset as it'll catch up eventually over a longer period. think Regis knows this though as we've clearly been taking a more offensive approach to the end of the season