Expected goals is a bad stat

how is expected goals calculated?

>Football data experts Opta have analysed over 300,000 shots to calculate the likelihood of an attempt being scored from a specific position on the pitch during a particular phase of play.
>The factors taken into account when assessing the quality of a chance include:

>Distance from goal
>Angle of the shot
>Did the chance fall at the player's feet or was it a header?
>Was it a one on one?
>What was the assist like? (eg long ball, cross, through ball, pull-back)
>In what passage of play did it happen? (eg open play, direct free-kick, corner kick)
>Has the player just beaten an opponent?
Is it a rebound?

Attached: Change.jpg (624x351, 20K)

You are wrong mate

oh so they just threw shit at tensorflow and hoped

How am I wrong?

>hoped
No, they analyse shots taken from every situation and looked at how often those shots resulted in a goal

What a fucking brainlet zoomer argument, nobody likes or accepts your bullshit made up stat lmao

Please explain to me how I'm wrong?

xG doesn’t show how good a player is. You don’t know how it works.

>you don't know how it works
It's not perfect, but given enough volume it works well. Look at Chelsea in 16/17. They won 17 more points than expected, Man City had the most expected points. What happened next season? Chelsea collapsed and City won the league.

xG it's just speculation
it's fake news

wow so if you're closer to the goal your chances of scoring go up. fuck me that's some revolutionary shit you got there mate. mind blown.

It is in football, seeing as anything deeper than goals and assists is a taboo to most fans

xstats are fucking retarded as it doesn't use any sort of context. I'd like to see soccer get more advanced with the way they keep track of the matches but this is not it.

So Higuain's and Messi's infamous WC final misses were from a zone that had a 0% chance of scoring

it's quite alright. not perfect but sufficiently good. if you compare actual points to xPTS in the bundesliga for instance you have an average deviation of roughly 4 points. i mean you decide if this is accurate or not.

For this to work shouldn't the position of every single player be taken into account too?

No?

If you have a defender right in front of you the probability of scoring will be way lower then if you don't

Feel like that stat doesn't take into account enough factors

>0-0 is the perfect result

Attached: B0CE512B-F877-48F1-834B-D810FEE94436.jpg (223x226, 10K)

Does each player have a different xG modifier?

If two different players have the ball 10 yards from goal; one is Lionel Messi and the other is a blind man with no legs; and they both shoot, does that result in the same xG?

>does that result in the same xG?
yes

Predicting the outcome for a specific singular situation isn't great for this reason, but it evens out with a bigger sample size. You can win one game when your xG is 0.1 and the opponent's is 3.5, but for an overwhelming majority of the time the team with 3.5xG is gonna fuck you up.

R E T A R D
A L E R T

>'09 Barcelona play 100 games against St. Swivens Catholic Girls School
>Barce being nice let the girls have 50 penalties a game
>They score none because they're shit little girls
>Barce limit themselves to 10 shots a game

Any games the little girls lose was just a fluke and I expect the girls to win the majority of the games because they'll have an xG of 80 and Barcelona can only have 10 xG maximum per game.
It'll even itself out over time.

>wow, xG is only valid when both teams are playing a serious game of football, what a shit stat

>that zoomer who never actually watches games but thinks they know more about football than everyone else after looking at a page of stats
This is the problem with stattos. They take stats as a replacement for actual experience, not an accompaniment.

stick to rugby

>goal is equally likely from penalty spot and 90 degree angle from five yards out
I don't think so.

Attached: arnie stop.jpg (194x259, 8K)

>Football data experts Opta

Attached: 1345562999344.jpg (155x202, 7K)

So every good team will just massively exceed their xG since their players score more often than random shitters?
And therefore xG is a terrible predictor of when a team is outperforming expectations, because there's no way to differentiate between "outperforming expectations" and "just better than other teams"?
What a shit stat.

But it's not valid then because it doesn't take into account the actually expected chance of scoring a goal.
You're saying that over time Joel Matip and Sergio Aguero would be expected to score the same amount of goals given the same opportunities.

>You're saying that over time Joel Matip and Sergio Aguero would be expected to score the same amount of goals given the same opportunities.
Epic strawman, no one has said anything that even remotely resembles that. Better finishers have better goals per xG, that's just basic analytics. A team with more goals than expected are relying on better finishers, even though the difference between goals scored and xG isn't very big even with the teams that have the best finishers in the world (2017-18 Real Madrid with CR7 even scored slightly less than expected over the course of the season). It gives much more insight to the team's play than something like goals scored or shots.

So you're saying it means nothing then since it isn't actually telling me the expected amount of goals

No, because that is exactly what it's doing.

okay dude

You can use poisson process or develop a new formula as they did:
oogoal.com

>Ronaldo's best G90/xG90 in the last 5 seasons (2014-15)
>1.14 expected goals per game
>1.39 goals per game
>scored 1.22 times more than expected

>Messi's best G90/xG90 in the last 5 seasons (2016-17)
>0.85 expected goals per game
>1.18 goals per game
>scored 1.39 times more than expected

>Giroud's best G90/xG90 in the last 5 seasons (2016-17)
>0.54 expected goals per game
>0.92 goals per game
>scored 1.70 times more than expected

Attached: legoat.jpg (670x476, 92K)

assists is the worst stat

Huh? If you shooting then you're probably going to have a defender in front of you regardless of what position you play

stats are worthless in football
prove me wrong

Football is not above statistics, just like every other sport

The main point of it is to quantify the quality of shots. You can have 35 shot attempts, even 35 shots on target, and still get less expected goals than a team who has shot less than five times, because they got higher quality looks.

That's actually pretty useful as a tool since so far it just had been muh shots, muh assists, muh goals.

"expected" might be the most retarded shit of all time

this
it's like these statbabbies got so deluded with their stat tracking that they decided to dismiss the actual game itself and just go off what their stats expected to happen.
Don't know why anyone would pay any attention to this bollocks, it's beyond retarded.

If you care for this you're literally not a fan of football

>muh beautiful game is above stats
cringe

Stats are fine, we actually show a bunch of them during the match you tard

Yes, surface level stats, like how many shots a team takes. A stat like expected goals is just a level deeper.

You can convince yourself it's deep all you want, it's literally retarded.
It's so retarded that I reckon the only reason you're putting any faith in it is because somewhere it supports some claim you've made because there's no other reason to be this retarded.
If you're so detached from the actual football you're now not even making stats on what is even happening in the actual game, you literally do not give a shit about the sport but just the numbers and the marketing. This is the ultimate stat for people who don't actually watch games, but either catch the highlights or tune in every now and again.

If it's so retarded as you say, why does it consistency show the the truth? I've showed examples before, like Chelsea getting extremely lucky in 16/17 while City actually had the most expected points, then next season Chelsea falls apart and City win the league. If it's as terrible as you say it doesn't make much sense that it consistency shows which teams are actually the best

I expect a goal m8

because football is not a game about stats. Only 1 stat actually matters, the goals.
Also correlation isn't causation, a lot of shit changes from Season to Season.
You really think all Chelsea did for an entire Season was "get lucky". Shit, Leicester won the League and even that wasn't getting lucky, you don't win an entire Season through just "getting lucky".
You can love your stats all you want, they won't tell you what will actually happen in a game, and that's the joy of football. Getting so enamoured with stats that you lose sight of all that completely just for shit you think will happen is just a testament you don't give a toss about the actual sport.
A sport can not be summed up in stats, ever.

>Shit, Leicester won the League and even that wasn't getting lucky,
Yeah, that's why they were still good the next season. You're just can't get over the fact that your sentimental sport can be broken down just like every other sport. Dortmund also outperformed expected goals to start the season, now they're falling apart. Liverpool were outperforming expected goals as well, now they're always slipping. You can shit on expected goals all you want but that doesn't change the fact that it continues to be correct

>The fact that your sentimental sport can be broken down just like every other sport.
Mate this is a pretty shitty defence mechanism. I literally don't care if you can break this sport down, it doesn't bother me a single bit because your stats will never be able to consistently be able to predict a game, or a Season, and football will still keep on being the same football I actually like to watch.

But I've already provided 3 examples of how it has

The fact you keep reusing the same 3 says a lot. Besides, the point was games and Seasons, your stats can't predict that, if they were so flawless we'd have predictions for next weeks games and next Seasons results. But we don't, because there are too many variables and you just end up looking backwards and using confirmation bias where possible to "prove" that your stats work.

I honestly don't understand why you'd sit and defend stats this long, what's your agenda here? Seems autistic

Because like I've said, you want football to be seen as some sort of pristine sport thats above the common sports that embrace stats when that simply isn't true.

You didn’t predict future behaviour. You took one even and a later (past) event and then fit xG into it.

And I've told you I literally don't care for that. I even said earlier no sport can be summed up in stats, none. I'm not creating a false standard here, it's you
You're also making up narratives in your head just to try disprove them.

Seriously, what is your agenda here? Why the autism? When I saw the flag I thought you were the Gypsy Ronaldo poster at first but given no mention of him either you've become more subtle or you're just this autistic

>no sport can be summed up in stats
Every sport can be summed up with stats

xg is garbage

deadspin.com/why-soccers-most-popular-advanced-stat-kind-of-sucks-1685563075

Trying to predict game results, goals scored, etc using statistics isn't anything new. Bookmakers have been successfully doing it for decades.
Now some new age football analytics website starts using that same metric but expresses it in whole numbers instead of odds or percentages, calls it "expected goals" and you're sucking their dick. We don't care. If you want to look into the future, maybe try to make some money placing betts, sure look at the xG or whatever. Leave the rest of us to peacefully enjoy the game and occasionally glance at the stats that represents what's actually happened on the field and not what may or may not happen

And if that's your view, you clearly don't actually like sports much, so why are you here?
You keep looking back but don't offer anything new, that's how useless you're being. Just like your stats.

>And if that's your view, you clearly don't actually like sports much
You have nothing to back that up

Pretty sure i covered that in another post, but we're going more in circles than before so I can't be bothered anymore.
Given how many points you willingly ignore, it seems you're not confident in the stats yourself but still want to argue it for whatever reason. And you're not going to budge so screw this shit

Don't be upset buddy, every old fuck has to come to terms with new technology eventually

who made that image? cristiano?

>replying to the retarded aussie
you fuckers got here today right?

I don't think it's him, it's someone else

>if they were so flawless we'd have predictions for next weeks games and next Seasons results
The stat literally tells you the quality of chances based on how similar chances have been converted. That's it. It's pretty fucking accurate and can explain a lot. It can't predict if players are going to be in bad/good form, no one has ever said that. It's an analytic guideline, not a crystal ball, and that's what it's always been. On average it's pretty fucking accurate and there just simply is correlation between goals scored and expected goals no matter how you look at it.