What is xG? | By The Numbers
Vložit
- čas přidán 30. 01. 2018
- xG, or Expected Goals, is the current metric of choice for the football stats community and has gained some wider traction with its appearance on the BBC’s Match of the Day, as well as plenty of comment from pundits and analysts who feel it’s either ruined or solved football. Here we explain what it all means.
Subscribe to Tifo Football at bit.ly/TifoSubscribe and follow Tifo Football elsewhere:
Website: tifofootball.com
Twitter: / tifofootball_
Facebook: / tifofootball
Instagram: / tifofootball_
Podcasts | iTunes
Tifo Football Podcast: bit.ly/TifoFootPod
This Football Life: bit.ly/ThisFootLife
Podcasts | SoundCloud
Tifo Football Podcast: / tifo-football-podcast
This Football Life: / thisfootballlife
Produced by Tifo Studios: Taking an illustrated look into the beautiful game. - Sport
Back when Lukaku was considered an elite striker
🤣🤣
One of very few football channels out there, which could seriously improve your football knowledge. Great video Tifo Football.
Always coming up with interesting stuff. Keep them coming as always
Suraj's Opinion is Better I don't know if in premier league talk a lot about XG, but I'm from Italy and I never heard it...
I just love you guys. I'm sending this video and your channel to my 12 year old niece and 14 year old nephew who both play for the NewCastle Jets Junior Teams (Australia), your videos are succinct and comprehensive, I reckon they'll make a big difference to young kids learning the more technical aspects of the sport.
Who came here after FM21 added xG?
FM calculates xG differently.
Very nice topic. One of the most discussed stuff by soccer analysts. Nice to see you explaining it
Your channel deserve more subs with this quality of video
Explain about VAR
Can anyone please help me understand how is xG(probability) greater than 1 for the Liverpool vs Swansea example ?
( how can the probability of any event be greater 100% ?)
Yes, it threw me off as well.
Thank you for this explanation! I've always known what it meant. What I didn't get was the actual number used. You've cleared that up.
Great video. Really clear explanation.
Cool vid like always, this is a stat that's thrown around a ton. Most (including me) don't really get it. Thanks for clearing it up.
Its definitely interesting. I can imagine clubs already plot xG in a normal distribution plot when looking at players. Harry Kane to the very right of the bell curve in the premier league - for example.
I recently plotted possession vs. xG against non top 6 teams for Liverpool as from the eye it looks they struggle breaking down buses. The plot confirmed the theory nicely.
🤦🏻♂️
Great simple and informative video on XG, as well as the things it does and does not do.
Never seen XG but I could imagine Andy Carroll’s
great explanation.
There is an error - xG is the probability of the goal, and not the likelihood of the goal. It's the likelihood of the shot.
P(Goal | Shot) = L(Shot | Goal)
So if I understand correctly then higher xg mean you have good midfield and playmaker and if you score more goals then your xg it mean you strikers did great job
Absolutely outstanding video.
Perfect exlpenation. Thanks so much
its pretty similar to shot conversion rate stat, both show how efficiant/inefficient a player is, or if you take the stat for all players of team, how efficiant/ inefficient a team is
Fascinating and clearly explained. Interesting that a player who was just good at getting in good positions and scoring a lot of close range goals perhaps wouldn’t have their ability reflected in goals to xG ratio?
Can you make a video on how good are Barca this season (stats & records) like you have with City?
Finally I understand xg. Thanks pal!
It’s none of the things above.
It’s just another bit of effort to convince American Networks that the game just make sense for American public just like VAR (Americanized) and make pundits look like they know what they’re talking about.
Be ready for 4 quarter games, Time outs, challenges by the managers, repeated plays prior to a foul and most importantly a non-continuous game clock so that commercials can be inserted.
The big clubs, the networks and FIFA are slowly killing the simplest and the most beautiful game in the world just to sell their product to the US market (and don’t get me wrong, American public is just bystanders here)
I thought I was going to miss Joe's voice,..Alex's explanation overwhelmed any tonal need
Best football page 💙
Wow I actually understood that malarkey. Well done
I friggin' love statistics in football and how they make you understand the game better
FIFA 22 brought me here
Hopefully with this video, you guys will start using xg more often in your future videos -particularly in the "by the numbers" series.
Off Topic but is their any chance of the home nations folding and being replaced with a team GB to compete on the international stage? Very Very risky topic but would love a cheeky chat on it ;) Especially considering most of wales players are English anyway
There is almost zero chance Scotland, Wales or Northern Ireland would agree to it. Since modern association football was invented in the British Isles, the first logical 'international' games were between England and Scotland, who while being within the same country obviously felt their distinct identities were enough to justify having separate teams. That's the origin of the Home Nations teams, and there is almost zero chance of undoing a tradition that old.
Seems to me like just limiting yourself for the sake of trying to be different. So much could be achieved with a national British football team. Tradition is there to be broken. I was born in Cardiff but would consider myself firmly British not welsh
You seem to be a minority then
After Romelo Lukaku's poor performance in the world cup, I'm sure he's now an amateur striker.😭
I really wish it had been named CQ (chance quality) rather than XG. I've talked to several (usually older fans) who dislike XG and a large part of it seems to be the name "expected goals". Many are sadly unable to get past the implication that goals are in anyway expected.
As a person who works with stats and data on a regular basis I think the backlash against XG is a result of most people having no idea how stats work. Just because the data says something about one specific event (for example a football match) doesn't mean that that event "deserved to" or "should have" gone a certain way.
For example, after the 2016 US Presidential Election, a lot of people concluded that Nate Silver's "fivethirtyeight" model was wrong, since it said Hillary Clinton had a sixty-something percent chance of winning, while, of course, Donald Trump won the election. Silver's model wasn't wrong. In fact he said Trump had a thirty-something percent chance of winning the election.
Your video is spot on. XG is a model, not an exact measurement. Therefore, it is more useful for long term predictions than making conclusions about one game. It says a lot more about the repeatability of a result. We can debate all day over who "deserved" to win the Swansea vs Liverpool match. XG doesn't say Liverpool deserved to win. It says that if that game were played over many many times, Liverpool would likely have ended up winning most of them, and it suggests that Liverpool's approach to winning football games combined with their talent is overall much more likely to earn results in the league than Swansea's approach combined with their talent (which is clearly seen by their respective spots in the league table).
I think people struggle with the concept that the model only becomes 100% accurate if simulated an infinite amount of times.
Not even then if, as stated in the video, it doesn't account for differences between individual players.
BigDave15
You can’t generalise that.
For example the model of fivetthirtyeight does exactly that.
I’m quoting their methodology now:
All players who have enough shots in our database to qualify are given a modifier based on their historical conversion rates (the number of goals they’ve actually scored, given the quality of the shots they’ve had). For example, Lionel Messi has historically converted a shot into a goal about 1.4 times as often as expected, so the probability of any shot he takes is multiplied by 1.4
How do I add xg and xga together?
Amazing video, as usual!
I think it's amazing how football stas are evolving on Europe, being heavily used by coaches and now even on TV, just like american's sports.
I also think it's amazing how little we hear from those stats here on Brazil teams and TV's... No news that we're still in the past =/
But I always see people say that player x is better than player y because he has a better expected goals i hope i get an explanation
This was a perfect explanation of what xG is.
did everyone get a black screen or is it just me?
Sheffield United's xG this season (12.03) is higher than Southampton's (10.93)
Sheffield United have scored 4 compared to Southampton's 19
Could u do a video on why arsenal defence is so porous inspite havin good defenders
They play too high on the pitch without a good Defensive Mid. Once their forwards press gets beaten, their Central Defenders are usually left on an island since their fullbacks are usually high up and are the ones who provide the width on attack.
And Arsenals defenders are good, but not great, nowhere near to the quality of City, Spurs, Utd, Chelsea, and nonexistant defensive midfield
I have a question. Can someone help? Let’s say Messi scores 10 Goals but has a xG of 5. Does it mean he was lucky? Does it mean hes really good because he scores against the odds? Or let’s say his xG is 15. Is he unlucky in that case? But a higher xG would be good because even though someone is underperforming it means that over time it is expected for him to score more often or not? So what exactly is good or bad regarding xG?
Great topic. I imagine xG is also useful from a Goalkeeper stand point, right? I'd like to see, for instance, a De Gea-Nick Pope comparison, to check who saves more goals than is expected, etc.
Theoretically. The problem with that is that the xG does *not* say anything about the quality of a shot, just how likely it is to go in under 100% average circumstances. For example a badly hit shot from the 6-yard with an xG value of 0.7 can be easier to save than a perfectly hit 20 yard screamer.
There are other metrics to determine goal keeper performance much better.
Good instinctive analysis..
It has really helped my fantasy football because I do not pay heed to it, and others take it as gospel. I look at body lanaguage and the manner in which that person hits the ball when they have a shot. Xg does not tell you that!
This channel is a voice of reason within an opinionated and biased community. Really impressed at how you guys can resolve the xG arguments in a 4 min video.
I find it very useful regarding predicting matches. I quite don't get why it shouldn't be. Especially if you look closely on the xG's without penalties and own goals and the difference in home or away performances.
There's so many external variables that not taken into consideration in this model. For example the quality of the goalkeeper. A striker will score more goals and have a higher goal return against a crap goalkeeper even in unfavourable situation with low expected goals. Similarly, a striker will have lower goal return against a world-class goalkeeper even in favourable situations with high expected goals. This model is pretty bullshit.
@@weixiangpeh6988 Of course that is a factor. But if you bet on matches you can sort that out. My point was related to the fact that through xG you can recognize if a team can create good chances on a regular.
Never forget Alex Stewart
So an xG of 1.0 doesn't exist and then Liverpool had an xG of 2.3? I am confused...
Yeah man, this video doesn’t explain it well.
@@CBM64 I rewatched now that you commented, so apparently they add up the xG of all the shots and come up with a total 2.3 xG for Liverpool during the whole match. Seems pretty obvious now buy I missed it before :D
@@Vox24 Ok, thanks that makes sense. And that seems to be what they do for xG for a whole season too, add it up. So you got some very high numbers in the end.
Which team has the highest xG on average? And which team has low xG and is "overperforming"?
City have the highest xG and Burnley are outperforming their xG and aG too.
City hve the highest, Burnley are overperforming by quite a bit as they do every season
+Harry Redknapp
Top, top, top, top reply that was, Arry. Very lucid than mine.
"You tried to hit it in the goal and you hit me... you've gotta work on your xG rating, son!"
"No wonder he's in the fackin' reserves."
hahahaha
Also, what about angle of pass? If it's a case of giving a specific xG rating for chances from a certain area, I don't think that's enough - it seems too simplistic. What about chances taken from that position, but one where the ball comes across the strikers path (making the margin for error smaller)? These are surely two very different chances. To bunch them all together based on absolute position on the field seems like something I shouldn't take seriously.
That's why, as shown in the video, different xG models incorporate different circumstances. When ball tracking is regularly used in the future, there'll be another layer to add to the metric.
Also, no one is telling you to take it seriously. This video just explains what it is.
Most models and all models used by major tv networks and clubs take those factors into account.
Only paying attention to the position the shot was taken is where the xG metric started, not where it is now. Nowadays good xG values consider many more factors than just the more position of the ball like the pressure the shooter faced, the game phase (counter attack, slow play etc.) or like you said, the position of the pass in relation to the position of the shooter.
back in the time when Lukaku and Sanchez scored more than expected :D
Lukaku actually still does.
Do gyan asmoah tacticas and top 5 African teams tactics as well
If xG is a probability or "chance quality" then why are the quoted numbers greater than 1?
Morata xg: 30 goals, actual goals: 10, yeah he was so worth the 70 million Chelsea paid, I hate football.
Football for dummies ! The best football channel for me!
I could see this being tweaked, to give goalkeepers an expected save per shot on target number, which would be way more insightful than the save percentage, and the saves per game-numbers we have today. What do you think?
Good idea! That metric already exists and is called after shot xG
My bad, post shot xG
if Xg is a number between 0 and 1 how can a team get an Xg of over 1
goals per chance?
What about the goalkeeper? If XG does not account for the goalkeeper then it is misleading. If the keeper is Courrtois, the XG will be higher.
Xg* Pg (for a particular striker) = real Xg
I thought this was pretty straightforward.. "Expected Goals" 🤷🏽♂
⭐️⭐️⭐️
Wait, if an xG of 1 is impossible, how did Liverpool had 2.31??
🤦🏾
Expected goal analysis is probably the worst model to analyse a player/team.
Why?
Swansea beat Liverpool then Arsenal next.
Why is the xG best used over periods of 5-10 games when it's more likely that the probability will be borne out over a larger sample? I don't really know how probability works, but that seems counter-intuitive on the face of it.
Also, how are players' xG rating over the course of a season determined? It seems to me that you can't really tell me how high an xG Lukaku will have without first telling me the expected chances created by supporting players, since the fomer is dependent on the latter. It seems like it'd be a ridiculously complex thing, and any unreliability about the chances Pogba, Martial and Mata will create will compound once it reaches the stage where we're predicting overall conversion rate of chances and giving an xG rating.
For example, how can Lukaku's xG rating for 2017-18 be determined when it's likely that Man Utd will fashion certain chances that Everton do not? Assuming the method to be trustworthy and that Lukaku's xG at Everton was reasonably well understood, how does that transfer to Man Utd? You may say that Lukaku puts away x number of chances created through a ball over the top for him to run across defenders and finish, but are Man Utd likely to execute such passes from deep? Isn't it more likely that we will work our way forward a bit more, and we'd be more in the dark about how Lukaku deals with these situations?
No one is predicting xG here, it's just a reflective stat. As said in the video, it's basically a numerical value for those games when one team attacks loads but doesn't score, and the other defends a lot, but nicks a goal.
However, even if someone were trying to predict the xG of a player before a game (which they're not), expected chances created by other players wouldn't matter as individual numbers because they are already reflected in the striker's xG rating from previous games. Goals don't come without chances, so if the xG number reflect the sort of chances the striker gets, then the created chances from other players are already implicit in the number.
The video doesn't say it's that useful, it just explains what it's for and how it can be useful in certain situations.
I love it when people rant about the xG value (which of course has its downside and points of criticism) when they don’t even understand the metric.
Do penalties get accounted for in this xG?
Christian pulisic please. Next
Your model is useless, if it cannot predict anything or contribute to that. Any model, any statistic, any figure.
xG is an offensive metric. So a corresponding defensive one would be needed on top of that.
you should picture of messi and then talked about lukaku . lol.
Therefore if the goals and xG doesn't match up, you can say the keeper did better than expected or the strikers did worse than expected, right?
Confused by chrismd story so am here
An imaginary stat 📊
FM21 players. Hello.
xG ìs not a real thing. Hope this helps.
Should rather be, what is a Gx (Granit xhaka) #Xhakaout
0:02 XG
Its because xg = bollocks
and what a useless factor
So boring.
what a garbage stat. way overused
SLiCe3000 if you say so
Only if you judge games based on xG alone.
+Suraj's Opinion is Better
I am talking here about people who jumped straight to xG maps in order conclude which team deserved to win the game.
xG still way better than comparing the numbers of shot ( on target or not) and possessions.
@Ter Van Ruth well 95 % of the time the team with the higher xG deserve the win , there is ofc some special cases like always
+Vauk God
Not 95%. False. How many times teams fail to execute the right pass, make the right decision etc. Look at Burnley vs Man United game for instance. Burnley had a higher xG although United had better chances. Because the United players were selfish and making terrible decisions, they didn't take shots that have higher xG. Build ups and screws ups can't be seen from this metric.
Its a bullshit stat
It is really not. But don't judge games solely based on it.
this is soccer not football
This is football not soccer😡