xG and finishing under ETH
- Thread starter Samid
- Start date
When it comes to finishing it will either come with time and confidence or through adding more clinical players. It's not really a worry. Our goal should be to develop a style by which controlling and dominating games comes naturally to us which we are obviously far off. I'd say that while there is improvement, we still are to loose in possesion and having certain players doesn't help (Bruno over or under hitting so many passes, Rashford being Rashford or Ronaldo being 37).
Also Rashford had so many easy chances to score. That's just human error/ form.
I think we have to think about personnel as mentioned earlier. We should be able to field at least 9 or 10 players who suit the 'philosophy' and we don't have that.
Also Rashford had so many easy chances to score. That's just human error/ form.
I think we have to think about personnel as mentioned earlier. We should be able to field at least 9 or 10 players who suit the 'philosophy' and we don't have that.
Do you seriously think this? You must have no faith in data science at all.
With its' huge sample size, which keeps growing, it is very informative actually.
That’s so funny when it happened, must be the only explanation.Rashfords shot when through on goal was so bad the referee was convinced it had to have had a deflection for a corner! There was no other explanation
You've summed it up nicely.
Rashford was fine early on with his shooting, albeit not perfect, but then the keeper made a couple of class saves and his confidence dropped and so did his finishing.
Staying off here gives a totally different impression of a lot of the matches. United dominated it was a 1-0 smashing. Just surprised it took as long as it did.
it's true there is a lot of variance in xG, but that doesn't mean it has nothing to do with performance
the players with the best finishing consistently out-perform their xG over long sample-sizes
Bruno's shots were all over the place, only Rashford's chances were solid and should have been put away. I believe it was the keeper's homework that make the difference. Take Rashford's 1st chance as example, Martial would have chip in, but Rashford, knowing his game, will hit it hard. The 2nd LHS low shot was another classic Rashford goal. Young would try to hit the top LHS corner, but Rashford always low. Anticipation + homework, make the difference.
I am reading this thread and I am amazed how many people have mentioned it so far
Forum discussions have reached a new level, and it's definitely not a good level. I bet our xG under Fergie was always very low, which would probably drive these hipsters mad regulary.
There is a big variance game to game for goals-xG, but on a team level it will even out. Individual players will outscore their xG but we don't have those top finishers.
Variance game to game for xG alone is more about actual performance of the team.
Coaching can improve the latter, and we're heading that direction, but there's not much we can do make our attackers more clinical.
I actually think those people who try to laugh off xG/stats in general are the hipsters these days. The rest have just accepted the football is changing, there are more ways to look at it than just team A scored more than team B.I am reading this thread and I am amazed how many people have mentioned it so far
Forum discussions have reached a new level, and it's definitely not a good level. I bet our xG under Fergie was always very low, which would probably drive these hipsters mad regulary.
Although I do believe the push in the media for xG is more to attract younger generations. I assume the people from older generation don't get it / don't want to understand it, or are simply not interested in it. The question is though, why would they spend time on a football forum, especially in xG threads.
How so? It’s a measure of the quality of a chance, based on distance from goal, defensive pressure, goalkeeper position, angle, body part used to take the shot and more depending on which model is in question. What about that do you object to?
Teams that consistently produce high xG and deny their opponents of xG consistently finish higher in the table. If I said that teams that create high quality chances and deny them from their opponents finish higher in the table would you find that objectionable?
It’s way more predictive of the outcome of games than possession, number of shots or whatever other straightforward statistic you might think of. Whilst there is variance over a game or two (because of player decisions and skill), in the long run, very few teams outperform what their xG stats would otherwise indicate.
It was “dreamt up” by an Opta analyst who now works for a Premier League club. Clubs now spend millions on data sources and analysis. At that level, they’re now looking at stats even more obscure than xG, like expected threat and possession value. If this field has no value, then why on earth do clubs invest so much into it?
I actually just watch all of our matches and witness it for myself, rather than reading stats.
That’s literally exactly what you’ve done.
Unless you can think of any other recent games with a similar pattern which makes you think last night’s game was a continuation of a theme?
Although, to be fair, you have identified a definite problem (we need a good striker) with an obvious solution (sign a good striker). We surely don’t need xG to work that out though?
Yeah but you could be a gormless moron for all we know.
So I think I'd rather trust the stats.
Rashfords shot when through on goal was so bad the referee was convinced it had to have had a deflection for a corner! There was no other explanation
Whilst being able to recall perfectly every single shot he’s taken through his career, whether he should or shouldn’t have scored and make an overall judgment based on that? The stats would suggest that over his career, he’s an average to slightly below average finisher, who finds himself in goal scoring situations more often than the average winger. Would you agree with that statement?
How does your statement refute what I've said? I don't need to be able to recall every single shot he's taken, I've watched 99.9% of the games he's played, and built an opinion over time. I haven't said he can't finish, he just misses quite a lot of gilt edge chances, which hurts us, considering how often he gets them.
That’s where xG is fairly useless. The eye test tells us that Rashford scores quite a lot of “low xG” chances. Picking the ball up a long way from goal, or smashing one in from range. But we also see that he has a habit of fluffing excellent “high xG” chances. These even out and the net result is a player who “matches his xG” yet this tells you nothing about the sort of player that he is.
Not sure that first line was necessary. Let him have his xG fun if that's his thing.
I wasn’t trying to refute your statement. The stats agree with you to some extent in this instance. My point was that we all have conscious and unconscious biases and unreliable memories, so simply relying on your gut feel and ignoring statistical evidence that may or may not agree with your instincts won’t give you the most balanced opinion. Statistics that challenge or support one’s opinions should be considered alongside the eye test.
I’d love to think these leeches will back him in January and let him get a striker.
There's a great book about heuristics and traps of what we think is being "logical" thinking, "Thinking, Fast and Slow" by Kahneman. The overall conclusion is people are really bad at estimating probability of sth happening and our intuition in terms of "stats" is quite terrible, we just feel like we're right because we base our estimates on what we remember (what is not much). We also remember things differently after a while.
It doesn't mean "an eye test" isn't useful, but it has to be put into some perspective and confronted with objective data (that equals to being more humble when it comes to your opinion).
Agreed. How you could watch Rashford every week and disagree with the statement that he misses a lot of good chances is beyond me.
My point was that stats like xG cannot be used as the sole basis for an opinion on a player's finishing, which is what was suggested by the other poster, who then classily decided to suggest I'm a gormless moron.
Shit happens. Their keeper had a great game. I'd rather be creating good chances and winning 1-0 than not creating any chances. Everton, we could have had more but we didn't actually under-perform our xG, it just feels like we did. We do need a quality striker though, but we've known that for ages now.
I mean that's not true and it's been repeatedly proven over the years. Ronaldo and Lewandowski are basically at their xG for the past 7 years, martial has always exceeded his xG. It's not about the finishing, it's about getting yourself in the high quality chances, repeatedly, and accumulating high xG.
Look at you swallowing the propaganda hook line and sinker.
Yep, I miss old days where people actually watched the match and not just follow stats and numbers like robots. Fecking modern football is driving me away more and more.
van Persie of 2012 would have had a hat trick and more on the chances created yesterday.
Ronaldo massively outperformed his xG at Madrid. Since he hit his 30s he hasn’t because he’s not as good as he used to be.
Yes, accumulating high xG is super important. But that is about who is the best goal-scorer and not who is the best finisher.
you’re basically trying to argue that everyone’s finishing ability is the same here
That just proves he is an average finisher, no? Top players outperform the statistical average finisher.
Look at you not engaging with the content of what someone says and instead hand-waving it away as “propaganda” because it doesn’t fit your pre-formed opinions. Strategy straight out of the flat-earther playbook there, well done.
Anyway @Pogue Mahone mentioned his perception of Rashford's finishing and that he misses more than his fair share of easy chances, even though his goals-xG looks average as he scores a fair few low probability belters. I thought this was worth examining as I would instinctively agree with the sentiment. The below table shows his performance vs xG with shots grouped into 0.1 xG ranges.
A couple of things to note:
- Understat.com data (only freely available source) so Premier League only.
- Understat.com's xG model is somewhat simplistic, it counts a shot and rebound as two distinct chances, more advanced models average the two (or more if there are multiple shots in the same phase of play). This might be advantageous as we are trying to assess his finishing rather than ability to get on the end of chances.
- I don't believe their model accounts for defensive pressure or goalkeeper location.
- Penalties have been excluded.
| xG decile | Total xG | Goals | Number of Shots | Goals-xG |
| 0-0.1 | 13.78 | 17 | 292 | 3.22 |
| 0.1-0.2 | 4.79 | 2 | 37 | -2.79 |
| 0.2-0.3 | 3.63 | 4 | 13 | 0.37 |
| 0.3-0.4 | 9.47 | 14 | 26 | 4.53 |
| 0.4-0.5 | 9.53 | 8 | 21 | -1.53 |
| 0.5-0.6 | 6.26 | 5 | 11 | -1.26 |
| 0.6-0.7 | 3.76 | 2 | 6 | -1.76 |
| 0.7-0.8 | 0.81 | 1 | 1 | 0.19 |
| 0.8-0.9 | 1.75 | 2 | 2 | 0.25 |
| 0.9-1 | 0.94 | 1 | 1 | 0.06 |
| TOTALS | 54.72 | 56 | 410 | 1.28 |
From what I can see, his overperformance on very low quality chances can clearly be seen in the 0-0.1 decile, and the sample size is fairly reasonable at 292 shots. For the level of variance involved it probably is a bit low, but short of having him take thousands of shots it should give us an indication. He does seem to underperform for chances in the range of 0.4-0.7 xG. These are very high quality chances, think a header from 6 yards on the low end and a central shot from around 6 yards with either foot on the high end. As a result any misses of this type would tend to be memorable, and will tend to stoke the perception that he is a poor finisher. That being said, the sample size is small and is completely balanced out by his overperformance in the 0.3-0.4 range, think central shots from around 12-15 yards.
| Type | Total xG | Goals | Number of Shots | Goals-xG |
| Head | 7.4 | 4 | 32 | -3.4 |
| Left | 9.55 | 8 | 51 | -1.55 |
| Right | 37.75 | 44 | 327 | 6.25 |
One thing that can be stated definitively, is that he is well below average at heading, which I think we can all agree on!
Ha.... keep it coming
Thanks for crunching the numbers. That’s interesting. The stats do sort of align with what I’m seeing. He has a habit of missing absolutely golden opportunities, balanced out by slotting more than his fare share of difficult ones. Although, yeah, that’s quite a simplistic summary!
The only year Ronaldo massively overperformed his xG was in 2014/15. Since the start of that season, he's scored 234 goals from 228.46 xG.
I'm not arguing that everyone's finishing is the same. I'm saying that the vast majority of clubs, players etc all eventually will perform roughly at their xG over time. Some, regularly overperform their xG. It doesn't make them great strikers. Martial has always overperformed xG for example, while Benzema, Ronaldo and Lewandowski since the start of 14/15 are basically dead on their xG.
This tells us that a great striker isn't defined by being clinical when they get a chance, but by getting a large amount of chances in the first place (which is both team play but far more the strikers own abilities). If a striker doesn't take a lot of shots or get a lot of chances, then he's a problem. If he's just missing chances, he'll probably sort it out and go on a hot run soon.
Best finisher vs best goalscorer... Best finisher is essentially a pointless and irrelevant debate. You don't want the best finisher up top. He might score 10 goals a season from an xG of 1. You want the best goal scorers, who will get 20+ xG every season and pretty much perform dead on their xG (if they do better, great, but not too important).
So as a club, if United is regularly getting high xG and temporarily underperforming it, it's really not a concern in the slightest. It's variance that will turn.
Not true. There's not a whole lot of correlation between the best strikers and outperforming their xG to be honest.
Benzema, Ronaldo, Lewandowski are all pretty much dead on their xG since 2014/15 for example, so I'd say that alone is a pretty massive indicator of that. Messi on the other hand blows past his xG.
Amazes me we still have some granddads that fail to give credence to xG, a completely proven metric that clubs themselves rely on. Apparently the infallible eye test is superior.
The eye test is usually distorted by whatever biases somebody has, and whatever they happen to remember. Many people don't even serve as useful witnesses to an event they are involved in, nevermind gain an accurate conception of fairly complex problems such as assessing the quality of chances over a whole game, or a period of games, something that requires analysis of a myriad of factors that make up a chance. xG is specifically created to capture this, the eye test is just a snap judgment.
There's a place for both because the eye test adds a lot of context but some just refuse to move with the times it seems.
The eye test is usually distorted by whatever biases somebody has, and whatever they happen to remember. Many people don't even serve as useful witnesses to an event they are involved in, nevermind gain an accurate conception of fairly complex problems such as assessing the quality of chances over a whole game, or a period of games, something that requires analysis of a myriad of factors that make up a chance. xG is specifically created to capture this, the eye test is just a snap judgment.
There's a place for both because the eye test adds a lot of context but some just refuse to move with the times it seems.
XG this season wouldn't really say that. If you were going purely on stats, guess you could point to territorial dominance stats vs turning that into meaningful xG, especially for the striker, then you could probably get to that answer (obviously we need a striker, just saying that wasn't a problem yesterday). Yesterday was purely on just missing shots on the day. Like our xG total yesterday wasn't far off from the xG we accumulated in the 9-0 vs Southampton. Just one day we scored 9, the next we scored 0... Combined xG from those 2 games was over 9, combined goals was 10... Just instead of a 5-0 and 4-0, it was a 9-0 and 1-0.
A serious weakness, our finishing is. One does not read to read the math to see what is obvious on the pitch.