How are the xg for and against game by game all through the season? Have we improved on our first three games?
Match Stats - Data Perspective
- Thread starter Borys
- Start date
For setting up a league table, points is the stat you want more than goals for and against. If you are interested in looking behind results, for instance to see developmental trends, you can either a) watch the actual games and form your opinions, b) check statistical measures that are relevant for how the team wants to play, or c) both of the above. I like looking beyond results, and I like option c) for that.
The work done by @Borys is of great enjoyment to me because of that.
Sorry I wasnt knocking anybody who wants to look at stats in depth, that wasnt my intention. Im just saying im 60 old fashioned and cant understand all the stats they come out with nowadays xg and stuff.
And I agree with that point, so I'm not really focusing on individual chances and how are they rated. It's not particularly interesting and highly affected by model and assumptions.
However, I think analyzing cummulated xG in individual games, and comparing it with number of shots gives a lot of insight how teams play. You can also see trends in how the team is performing over time. This was why I set up this thread, seeing a lot of positives after WBA game, which was far from general opinion on here. Seems I was right to expect much better results as we're not that bad recently.
Some people are obsessed with micro-analysis second by second how we conceded 2nd Leicester goal. For me it is totally pointless as there are million possible scenarios. What's interesting, defensive side of our performance vs Leicester was very very good judging by the xG and number of shots. It is a different perspective, that's all.
Edit: by the way, I guess there are people on here who know more about this stuff (fantasy football players etc), but if you want to get the basics I think I explained it in a easy way on the first page.
The time graph is a must, much more information in that. That's why I'm using understat. I suppose they do it manually as well?@Borys Here's a screenshot of the match report from xgstats.com where every shot is watched and recorded manually (by a human).
I think that's more in line with reality having watched the match.
Last edited:
I think it's just one guy watching every single match in from Premier League to National league and recording every single shot (I have a mastersheet with all 25000 shots taken so far this season). Maybe he's hired more people now I don't know
But can't knock his accuracy. The number of goals scored match his xG numbers almost 100%.
Borys, thank you for all your efforts in here. Hopefully one of the United staff that roam this forum can put in a good word for you in the data analysis department. You deserve it.
Overall, it looks like we are dominating most games in almost every phase of play. Long may it continue. Ole has developed a strong foundation at this club.
Overall, it looks like we are dominating most games in almost every phase of play. Long may it continue. Ole has developed a strong foundation at this club.
This is what is so commonly wrong with people's interpretation of stats. They fear what they don't understand and distill it down to something that they can. I mean look, you mentioned a quote about possession from SAF that has nothing to do with xG.
xG stats does not have a 1.0 correlation to match result - after all, football has a lot more luck and randomness involved by nature. However, on a large scheme analysis it's an absolutely pivotal variable that is key to analyze. Why fear the unknown that can help growth?
Thanks a lot!
The bolded part is very important - on a large scheme is key. No point in focusing on micro-events, like we've been discussing earlier.
Fully accapted, wasn’t that harshly intended. There are many ways to enjoy football (or anguish about it, equally important!)
https://fbref.com/en/squads/1953887...ter-United-Scores-and-Fixtures-Premier-League
Statsbomb's model (which fbref uses) is considered to be significantly more advanced than Understat's but here's Understat's United section:
https://understat.com/team/Manchester_United/2020
To answer your question, the first 3 games were awful, especially the Spurs match. Since then generally fine though quite a few close games on xG.
The xG last season was interesting. We look comfortable winners of the first 6 games yet in reality only won two of them. https://fbref.com/en/squads/1953887...ter-United-Scores-and-Fixtures-Premier-League
Fernandes turned a team that was creating chances but unable to take enough of them, into a side that regularly won matches. While on the defensive side, we were somewhat unfortunate in the first half of last season but benefitted from opponents missing chances in the 2nd half of the season.
Last edited:
Really excellent work Borys. Every now and again I struggle to sleep at night and will search for things to read. Definitely, this is one of the more insightful threads I have seen recently.
I looked at the fbref site and did a quick sum of expected points (pl season so far) based on XG and XGA and compared us to our rivals i.e, Liverpool, Man City and Chelsea. What's interesting is that we've outperformed our expected points while our rivals have all underperformed against it. The only conclusion I can draw from this is, while we are progressing, we are still behind those other teams and the fear is that eventually the table will reflect this.
I looked at the fbref site and did a quick sum of expected points (pl season so far) based on XG and XGA and compared us to our rivals i.e, Liverpool, Man City and Chelsea. What's interesting is that we've outperformed our expected points while our rivals have all underperformed against it. The only conclusion I can draw from this is, while we are progressing, we are still behind those other teams and the fear is that eventually the table will reflect this.
To answer that question, I updated xG graphs. For United there is a huge improvement after the first 3 games:
My general understanding is that if your diff xG (xG-xGA) is below 0,8, it's most likely to be a draw. Now looking at that graph we can conclude that:
- we were a bit unlucky to lose against Crystal Palace*
- we were extremely lucky to win against Brighton* (obviously)
- we were a bit lucky in West Ham* game as they created a lot of chances to score (2,2 is very high xGA)
- we did well to win against Sheffield* as again they've had some quality chances (xGA 1,6)
- we were unlucky not to win against Leicester with xG diff 1,0
- in other games xG diff corresponds well to the final result
- in games vs Arsenal, City and Chelsea our approach was very cautious (targetting low xG/xGA). It would probably end in 0-0 for each game if it wasn't for Pogba conceding a penalty vs Arsenal*. Leicester was the first game against top opponent this season we really went for the win.
- just by looking at the lower graph, you can see we're creating a lot (certainly enough) of chances every game. Our issue is it goes with high xGA and we also concede a lot (surprise surprise).
Let's take a look at xG and GD charts for all title contenders below:
Conclusions:
- Liverpool were unlucky to draw against WBA and Everton, lucky to win vs Wolves and to draw against Brighton, lucky to win against Tottenham. In other games xG correlates well to final result.
- Chelsea definitely look like underperforming considering their xG diff
- Tottenham expected results are far from final result. It's probably because Mourinho is a weird guy and highly relies on indivudials to win him games. They did ride their luck a bit earlier, somehow unlucky in a couple of games lately. I'd say neutral. They the only low xG team from all contenders, which is typical for Mourinho teams.
- City results correspond well with xG, with a couple of exceptions (Spurs obviously). Their xGA is pretty amazing, I'm still trying to understand why they concede so little chances.
Last edited:
Really good thread @Borys , best thread in the football forums in Caf. Kudos to your incredible job here.
Checking at the xG it is a bit worrying. We are doing a bit better than the xG-xGA seem to suggest. For example, if we round xG-xGA and give point, we should have 27 points (so, we are better than we should be). I don’t know what does this mean, an inevitable return to the mean soon enough, or just a clinical ability of our forwards. We probably won 3 matches which we probably should have not won (including the second match of the season which we should have lost), and on the other hand, lost 2 points (which we should have not) against Leicester.
The main important thing is that those metrics seem to be improving (same as the results) which show that we are in the right track.
Checking at the xG it is a bit worrying. We are doing a bit better than the xG-xGA seem to suggest. For example, if we round xG-xGA and give point, we should have 27 points (so, we are better than we should be). I don’t know what does this mean, an inevitable return to the mean soon enough, or just a clinical ability of our forwards. We probably won 3 matches which we probably should have not won (including the second match of the season which we should have lost), and on the other hand, lost 2 points (which we should have not) against Leicester.
The main important thing is that those metrics seem to be improving (same as the results) which show that we are in the right track.
To answer that question, I updated xG graphs. For United there is a huge improvement after the first 3 games:
My general understanding is that if your diff xG (xG-xGA) is below 0,8, it's most likely to be a draw. Now looking at that graph we can conclude that:
*in alll those games Pogba started in midfield. Stats don't lie.
- we were a bit unlucky to lose against Crystal Palace*
- we were extremely lucky to win against Brighton* (obviously)
- we were a bit lucky in West Ham* game as they created a lot of chances to score (2,2 is very high xGA)
- we did well to win against Sheffield* as again they've had some quality chances (xGA 1,6)
- we were unlucky not to win against Leicester with xG diff 1,0
- in other games xG diff corresponds well to the final result
- in games vs Arsenal, City and Chelsea our approach was very cautious (targetting low xG/xGA). It would probably end in 0-0 for each game if it wasn't for Pogba conceding a penalty vs Arsenal*. Leicester was the first game against top opponent this season we really went for the win.
- just by looking at the lower graph, you can see we're creating a lot (certainly enough) of chances every game. Our issue is it goes with high xGA and we also concede a lot (surprise surprise).
Let's take a look at xG and GD charts for all title contenders below:
Conclusions:
Overall I'd say United and Liverpool have been slighlty better than expected, Chelsea significantly worse, and City&Spurs just as expected. However, as I expect Chelsea to improve, Liverpool tend to get results even if xG diff is against them, while for us the mentality has changed (in a good way ). Also, we're still struggling with converting easy chances and I don't expect De Gea to be this poor for the entire season.
- Liverpool were unlucky to draw against WBA and Everton, lucky to win vs Wolves and to draw against Brighton, lucky to win against Tottenham. In other games xG correlates well to final result.
- Chelsea definitely look like underperforming considering their xG diff
- Tottenham expected results are far from final result. It's probably because Mourinho is a weird guy and highly relies on indivudials to win him games. They did ride their luck a bit earlier, somehow unlucky in a couple of games lately. I'd say neutral. They the only low xG team from all contenders, which is typical for Mourinho teams.
- City results correspond well with xG, with a couple of exceptions (Spurs obviously). Their xGA is pretty amazing, I'm still trying to understand why they concede so little chances.
My main takes from this is that it confirms the eye+memory test for me, which is a good thing:
- First three games were an anomaly to be ignored mostly.
- United are now consistently one of the three/four best teams in general play, and we are catching up on Liverpool and City bigtime.
- We have been steadily improving for more than a year, since before Bruno.
- City and Liverpool should still be favorites for the title, but we are now a serious dark horse.
Manchester United 2 - 1 Aston Villa (Martial, Bruno Pen; Traore)
xG=1,9 xGA=1,6
What happened in the second half? We kept Villa at bay for the first 45mins. After the break we totally lost momentum: McTominay losing the ball due to an easy error, Fred losing the ball with a poor pass, Maguire fouling Grealish (pulled out of defense), Shaw with a heavy touch. We looked threatening, but couldn't keep the ball at all. Villa looked confident and found themselves in a couple of promising situations early on, which resulted in two CLEAR chances (better than our chances) and scored one. It all happened between 45-60th minute of the game. We got a penalty and then had another big chance but overall I think it was a pretty close game between two in-form teams.
Anyway, the question remains: what changed in the second half, was it something in AV setup, or maybe we were too casual? Sometimes I think we are not aggresive enough to close down players, but this is just a subjective opinion and I might be wrong. Yesterday I saw this a few times with AWB and McTominay, just standing in front of Grealish waiting for his move. I think in that situation they should at least fake a tackle or challenge him physically.
Some other points:
Overall I thought the first 45' were the highest level of performance this season, considering Villa are no pushovers. Second 45' we lost momentum, I don't quite understand what happened (fatigue? lost focus? to passive off the ball?), and we did well to get 3 points from that game. However, Bailly, Rashford and Pogba are hitting good form, even Martial looked quite sharp so this is the right time to face City and Liverpool. But I'm glad boys will have 5 days rest now, deserved after busy festive season (from which we came on top by the way). Onwards and upwards, some big positives, and still a lot to be improved.
xG=1,9 xGA=1,6
- possession 47%
- shots on target 8/18 vs 5/15
- 4/5 shots saved - signs of improvement from De Gea?
What happened in the second half? We kept Villa at bay for the first 45mins. After the break we totally lost momentum: McTominay losing the ball due to an easy error, Fred losing the ball with a poor pass, Maguire fouling Grealish (pulled out of defense), Shaw with a heavy touch. We looked threatening, but couldn't keep the ball at all. Villa looked confident and found themselves in a couple of promising situations early on, which resulted in two CLEAR chances (better than our chances) and scored one. It all happened between 45-60th minute of the game. We got a penalty and then had another big chance but overall I think it was a pretty close game between two in-form teams.
Anyway, the question remains: what changed in the second half, was it something in AV setup, or maybe we were too casual? Sometimes I think we are not aggresive enough to close down players, but this is just a subjective opinion and I might be wrong. Yesterday I saw this a few times with AWB and McTominay, just standing in front of Grealish waiting for his move. I think in that situation they should at least fake a tackle or challenge him physically.
Some other points:
- heatmaps for this game are... inconclusive. It doesn't look like a diamond, because nobody played the "defensive tip" (Fred roaming around, McTominay sticking to right hand side). I'd say it was a 4-2-2-2 with Pogba and Fernandes roaming.
- possible reason for our poor second-half spell was McTominay very poor game: 74% passing accuracy (which is very low for him, and I don't even think Villa pressed him much), disspossed 4x (top), lowest % of successful presses (1/18), only 2 defensive actions (tackles + interceptions), lost 3/5 headers, joined max number of dribbled past (2x) and he only played 60'. He looked a bit out of place which in my opinion happens every time we don't play flat 4-2-3-1. As a big fan of Matic, I hope he gets Scott place in first XI if Ole decides to go with any more fancy formation.
- watching the game I thought it was an excellent game from Pogba (and I say this as one of his biggest critics). Stats confirm that: level with Fred on tackles+interceptions (7), but more clearances (3 vs 1), 3rd most succesful presses (4/11- hard to say if that's a high number). What's interesting he had only 1 dribble, which means he relied more on his passing than quick feet (at which he is better in his head than in reality). His passing accuracy wasn't very high which is not a big deal as he was high up the pitch. He matched Bruno in shot+goals creating actions. Also won 5/6 headers. Personally I hope we never play him in midfield again, considering how good he was with a little more freedom and no specified role in terms of positioning.
- Fred and Pogba were our most involved players in terms of passing, which is important because the general trend is our CBs top that parameter every game. Probably (might be wrong here) it is caused by intruduction of Bailly, who doesn't pass it back to GK (in which case ball goes back to CBs and their passing numbers are pumped.
- special mention to Bailly performance. 90% passing accuracy, most interceptions (9) and clearances (5). What stats don't show is how good he is at starting attacks. Another interesting stat is Eric is the only player from whole defense + midfield who didn't attempt a single tackle, and yet was our most important defender.
- I though Rashford was very good. I believe he's actually more comforteable on the right. Decision making has to be better but his dribbling has improved a lot lately (4/7 successful dribbles), in general he doesn't lose the ball due to poor touch and creates decent shooting chances for himself. His shooting technique is also good, the ball is moving almost straight so he I'd make that swith to right side permanent.
Overall I thought the first 45' were the highest level of performance this season, considering Villa are no pushovers. Second 45' we lost momentum, I don't quite understand what happened (fatigue? lost focus? to passive off the ball?), and we did well to get 3 points from that game. However, Bailly, Rashford and Pogba are hitting good form, even Martial looked quite sharp so this is the right time to face City and Liverpool. But I'm glad boys will have 5 days rest now, deserved after busy festive season (from which we came on top by the way). Onwards and upwards, some big positives, and still a lot to be improved.
Last edited:
Villa’s PSxG was 1.4 so that does suggest some improvement from DDG, although any PL keeper should have saved the shots he faced last night in my opinion. I think the goal would have had a pretty high PSxG.
Very good analysis, as usual, thank you for sharing with us!
McTominay was tasked to mark Grealish, in my view. And he was dragged out of position many times. I think Ole decided this is not working out and that's why he subbed him off for Matic, who then played positionally, not man-marking.
Look at Grealish's heatmap and compare to Scott's:
On Rashford, it definitely helps the team because he keeps the width, combines well with Wan-Bissaka and is threatening on a dribble. Hope he is played there more. It could also help his development, appreciating and adding elements to his style.
3 games in 6 days isn't it? I think we might be able to put it down to fatigue. I share the same sentiments, first 45 was very good, second 45, not as much. In the first half it was mainly just McTominay playing poorly but Fred was at his usual high level, but in the second half Fred decided he wanted to start playing poorly too, when 2/4 midfielders (especially the two more defensive minded ones) are playing like they were doing it cedes control of the match. The diamond relies on all 4 of your midfield players being on it all game, like we saw against Arsenal, if your midfield has an off game then the diamond loses effectiveness, towards the end of the game it looked like we switched back to the 4-2-3-1 which we're more comfortable with. Personally I'm not keen on the diamond experiment, perhaps it's only a case of personnel that we're using it semi-frequently. I still think the lack of a RW is really limiting what sort of game we can play. Excellent thread by the way @Borys
Ole reacted quickly by replacing McTominay with Matic. However, it seemed unplanned as Nemanja main objective was to kick football out of Grealish mind. But he helped us to get some balance back as well.Very good analysis, as usual, thank you for sharing with us!
McTominay was tasked to mark Grealish, in my view. And he was dragged out of position many times. I think Ole decided this is not working out and that's why he subbed him off for Matic, who then played positionally, not man-marking.
Look at Grealish's heatmap and compare to Scott's:
On Rashford, it definitely helps the team because he keeps the width, combines well with Wan-Bissaka and is threatening on a dribble. Hope he is played there more. It could also help his development, appreciating and adding elements to his style.
Next time we want to mark somebody I guess Fred would be a better option, he is comfortable in all areas while Scott has a strong preference for right side.
I forgot to add another point I've considered recently. I believe we are a bit too offensive at times, too direct. What could help us regain some balance is to play some passing game between the defenders, midfielders, around the box and back to keeper. I know fans want to see attacking football but at times it seems like we any sense and play high risk high reward all the time. We should focus on keeping possession and regaining momentum and shape at times. Liverpool and City do that, and I believe it could take us up a level. We have the players to go that, it's a matter of approach.
Agree, we are a very 'route one' kind of team. There is only one thing on our players' mind when they have the ball. We only recycle possession when we are comfortably in the opponent's half and are besieging them.
Cheers mate.
I think the Aston Villa game was really good and show that we mean business if we can perform the same week in week out with more clinical in the final third. The front four Bruno, Pogba, Rashford & Martial were very fluid & constantly moving around which made Aston Villa players difficult to know who they are marking.
McTominay was the only our bad player in that match from what I watched which backed it up with your stats that he was so sloppy on possession even without being pressed. I can forgive his poor possession ball game if he did very well to win it back and also defensively but he wasn't effective enough defensively in that match either.
Rashford showed good work rate, good link up play to build up the play, had some shots with both his feet were saved by Martinez. Martial wasn't just scoring goal but also the link for oue front four & Pogba was constantly everywhere in both offensive & defensively.
Based on your stats, it shows that Pogba will perform better if he can just cut off his dribbling & too many touches and focus more on his passing. And I think he needs to improve more on his finishing.
I think today's game show that if we want to reach the league title, we need to have centre back like Bailly but can stay fit week in week out to pair with Maguire, pace, quick to react, bravery & very good one on one. We also need to upgrade McTominay to someone who is equally physical presence like him with equal work rate defensively but better on the ball and possession. And it's becoming clear if we lose Pogba in the summer means we must replace him (no 1 priority) with someone who is creative player to replace his creativity which Grealish is a good shout for candidate.
It certainly looks as if Bailly is having good effect on our back 4. Also, I can see why people are so obsessed with getting new DM. McTominay has done well so far but but he prefers action to be in front of him, rather than being in the middle of action. He is a defender playing in midfield.
Anyway, what we've learned so far is we rarely struggle to create goal scoring chances, but also we concede a lot of chances compared to our rivals. Now I don't think this is down to personnel, it must be something with the system. How come City are getting so low xGA every game? Are they more aggresive around the box, or is is because they keep the ball better (what we've discussed in previous posts)?
If we can improve that, we might as well be real title contenders. We are not just yet, but we're doing really good to keep up with Liverpool. I think they are comfortably the best team because they can cope without a few key players injured, a benefit of the system they play but at the same time it looks like this system is quite energy-consuming so it'll be interesting if they feel fatigue earlier than some other teams in this weird season.
I think it's down to Pep's system. They keep the ball away from you and cut out counters as they start, often through nefarious means.
City's xGA is usually been significantly lower than every other team in the league under Pep. Last years ~39 for the season was their highest since Pep arrived. They're on course to be lower than last year but not as low as 2017/2018 when it was around 25 for the season.
I don't have any stats to back this up but they seem to be playing more conservatively this year. Whether that's because Aguero's been out, their other strikers aren't firing or if it's tactical due to the congested season I'm not sure.
I don't think we've been doing too badly since Tottenham (We're third for xGA since then according to understat) but we seem to be giving up big chances, mostly through individual errors, and the opposition have been taking them.
Last edited:
One thing that needs to be kept in the forefront when discussing xG is that it's calculated based on averages for the whole league.
Better-than-average players will exceed the xG for their attempted shots over time while less-than-average players will fall below the xG for their attempted shots. The same applies to teams.
I would also posit that better players and teams should be taking higher xG shots, as well as making more of them, than lesser players and teams. Scouts: Want to improve your team? Find players that take higher xG shots and make or exceed xG.
United SHOULD be exceeding xG. If we're not, we need better players.
And when someone mentions "reverting to the mean," teams should not be reverting to the xG unless they are average teams. Above-average teams that dip should revert to being above average. If they don't, something has happened that needs to be addressed. Below-average teams that rise have either addressed their problems to get there or they'll revert back down to where they were.
Better-than-average players will exceed the xG for their attempted shots over time while less-than-average players will fall below the xG for their attempted shots. The same applies to teams.
I would also posit that better players and teams should be taking higher xG shots, as well as making more of them, than lesser players and teams. Scouts: Want to improve your team? Find players that take higher xG shots and make or exceed xG.
United SHOULD be exceeding xG. If we're not, we need better players.
And when someone mentions "reverting to the mean," teams should not be reverting to the xG unless they are average teams. Above-average teams that dip should revert to being above average. If they don't, something has happened that needs to be addressed. Below-average teams that rise have either addressed their problems to get there or they'll revert back down to where they were.
Definitely their pressing game.
They aim to win the ball before it even reaches the opponents box, if you get through that then they've got Ederson sweeping.
Usually if you manage to beat their press effectively you've got a clear chance on goal, but beating the press is the hard part as they'll either win the ball or foul you in the process and regroup.
We at united aren't great at pressing yet, we often end up sitting back in a tighter formation and let the ball go wide - this is an issue though as we often let crosses come into the box far too easily (see aston villa game) which gives opponents a free chance at goal.
It works against some teams but others with good crossing wingers or fullbacks it can become an issue.
Is it average calculated for each league or in general between top leagues?
You mean scoring more goals than expected if I understand correctly?
To be honest I don't take this "total xG vs goals scored" for each team very seriuosly (as a cummulative values). This is highly affected by some odd games, I'd rather focus on analysing xG difference in relation to result (Goal Difference) for each game individually.
It does give you some important information on how good are strikers, because you can see the expected goals and total tally for each season. Some players have good numbers because they play for teams that create a lot, and some will be exceptional goal-scorers (like Martial and Greenwood for us).
Yeah I agree with that, we seem to be happy with opposition having the ball around the box. Aston Villa had a number of easy crosses, especially in first half. This is something I don't like about our game, we are not aggresive enough (and obviously far away from well-organized pressing, but improving). Wan Bissaka is a good example of that. Unless opponent doesn't try to go past him, he will not tackle - as a result, he is pushed deeper and deeper towards the goal.
I also think Man City are playing similar to us this season, with 2 midfielders + De Bruyne, which probably helps them to be more sound defensively, but I've not seen much of them lately.
xG has been calculated across leagues. The value of a penalty is going to be same across every league from Premier league to Estonian 2nd division, ditto for a shot from 6 yard box.
Not really, better players manage to get into better positions, but you'd be surprised to see the number of top players who consistently overachieve their xG by a significant amount. Last season was the 1st time since 2014/15 season that CR7 exceeded his xG. Lewandowski has underachieved his xG in 3 of last 6 completed seasons. Leave Messi out (who exceeds his xG because of ridiculous shooting from outside the box), and you'd struggle to find a top player get xG+3 to xG+5 at the end of a season.
CUP GAMES QUICK SUMMARY - no detailed data available
Manchester United 0-2 Manchester City (Stones, Fernandinho)
Our confidence was shattered after the first half, we've conceded (doesn't matter how) and City decided to sit deep and wait for a chance to counter attack.
I've made some calculations in another thread showing that contrary to popular opinion, beating smaller sides doesn't seem to be a problem for us since Bruno joined. I think we topped that table of pts/game when compared with other contenders, but it was calculated before festive period so it'll be outdated (and in fact, we've probably strengthened our position by now). What's more important, I pointed out we're getting significantly less points against big sides this season - now a widely discussed topic on here.
Anyway, no accurate data is available for cup games so I'll just write down my opinion on what we're doing wrong. Personally I think midfield is our biggest issue - but it's not entirely down to personnel. We are a top heavy team, meaning we play with two forwards + two AMs roaming (best explanation I've seen) when using 4-2-2-2 formation. The problem it creates is that we have back 4, two midfielders who have limited role in the buildup by default (we'll get back to that) and 4 players high up the pitch. Now this makes it very difficult to keep possession, because a) Fred and McTominay are not good at that and b) there are some serious issues how this team is coached in terms of moving the ball. I like what @tjb has pointed out in post-ManCity thread: " our CM's position themselves away from possession and don't have the composure to move the ball in one touch (...) forcing us to build up play slowly despite the quick movements of our attackers."
The bolded part is very important, just notice how big are distances between players during the build up - this doesn't seem to be a personnel issue to me, because even if we have well positioned midfielder we very rarely pass the ball to him, we always seem to go either through the wings or directly to attack. This has been an effective tactics in some games but it's not enough against organized teams who are good at pressing.
Regarding the composure issue, there are two players in our squad who are comforteable at quick short passing in tight areas, difficult angles and good against pressing: Matic and van de Beek. Every other midfielder is either scared of receiving the ball under pressure, or prone to simple technical mistakes (miscontrol). It is quite telling that those two are down the pecking order in games against the big boys.
City build up play from the back was quite ridiculously good really, they dragged our players all over the pitch to create space in central areas and then progress with the ball forward- at some point I thought we should stop pressing them because it's doing more bad than good for us. This is the exact opposite to what we do, meaning playing the ball between back 4 to find a direct pass forward, omitting midfield - in that case we rely on their individual brilliance, which simply wasn't there against City.
In my personal opinion, it's a basic mistake to go with midfield duo against pep teams. Obviously, it's hard to tell if we could win that game against City with different setup, but I think in general we need to move in the direction of midfield 3 in those big games to get more control of the game. I've been advocating using more of Matic and van de Beek for a while now, so far Ole seems to have another idea. Fred and McTominay are fine in most games, because they work hard to get the ball back in midfield and catch the opposition in transition. This seems to be working quite fine for us, but it didn't seem to do anything to City (88% of passing accuracy).
For front players, it's just very difficult to perform if they know they'll get the ball only a couple of times in a game. For anyone who played football it's obvious how hard it is to get into a rhytm in that kind of game when you're having 30% possession, so I wouldn't blame forwards much for underperforming in that game.
Our record against top teams is abysmal this season, this might have insignificant impact on title race in the EPL, but we're not winning any Cup playing that way. We need to re-construct the midfield not only in terms of personnel but also in terms of strategy how to move the ball and get some control in the big games.
PS
Why do I think this is mainly a coaching isse? Because I see a players in good positions being ignored, because players who can do much more are just passing it backwards (and you can see this is the only thing on their mind), and because we seem the be playing very safe most of the time, while more often than not good things happen when we speed it up a little bit.
We only seem to play triangles with a few players like Bruno, Mata, van de Beek. Most of the time the passing is vertical or horizontal, just basic stuff.
Our rivals are less afraid of losing the ball than we do, and take more risks. We can do that as well, even with current personnel.
Manchester United 0-2 Manchester City (Stones, Fernandinho)
- 40% possession (31% first half)
- 85% passing accuracy (vs 88%)
- 2/11 shots on target (vs 4/12)
Our confidence was shattered after the first half, we've conceded (doesn't matter how) and City decided to sit deep and wait for a chance to counter attack.
I've made some calculations in another thread showing that contrary to popular opinion, beating smaller sides doesn't seem to be a problem for us since Bruno joined. I think we topped that table of pts/game when compared with other contenders, but it was calculated before festive period so it'll be outdated (and in fact, we've probably strengthened our position by now). What's more important, I pointed out we're getting significantly less points against big sides this season - now a widely discussed topic on here.
Anyway, no accurate data is available for cup games so I'll just write down my opinion on what we're doing wrong. Personally I think midfield is our biggest issue - but it's not entirely down to personnel. We are a top heavy team, meaning we play with two forwards + two AMs roaming (best explanation I've seen) when using 4-2-2-2 formation. The problem it creates is that we have back 4, two midfielders who have limited role in the buildup by default (we'll get back to that) and 4 players high up the pitch. Now this makes it very difficult to keep possession, because a) Fred and McTominay are not good at that and b) there are some serious issues how this team is coached in terms of moving the ball. I like what @tjb has pointed out in post-ManCity thread: " our CM's position themselves away from possession and don't have the composure to move the ball in one touch (...) forcing us to build up play slowly despite the quick movements of our attackers."
The bolded part is very important, just notice how big are distances between players during the build up - this doesn't seem to be a personnel issue to me, because even if we have well positioned midfielder we very rarely pass the ball to him, we always seem to go either through the wings or directly to attack. This has been an effective tactics in some games but it's not enough against organized teams who are good at pressing.
Regarding the composure issue, there are two players in our squad who are comforteable at quick short passing in tight areas, difficult angles and good against pressing: Matic and van de Beek. Every other midfielder is either scared of receiving the ball under pressure, or prone to simple technical mistakes (miscontrol). It is quite telling that those two are down the pecking order in games against the big boys.
City build up play from the back was quite ridiculously good really, they dragged our players all over the pitch to create space in central areas and then progress with the ball forward- at some point I thought we should stop pressing them because it's doing more bad than good for us. This is the exact opposite to what we do, meaning playing the ball between back 4 to find a direct pass forward, omitting midfield - in that case we rely on their individual brilliance, which simply wasn't there against City.
In my personal opinion, it's a basic mistake to go with midfield duo against pep teams. Obviously, it's hard to tell if we could win that game against City with different setup, but I think in general we need to move in the direction of midfield 3 in those big games to get more control of the game. I've been advocating using more of Matic and van de Beek for a while now, so far Ole seems to have another idea. Fred and McTominay are fine in most games, because they work hard to get the ball back in midfield and catch the opposition in transition. This seems to be working quite fine for us, but it didn't seem to do anything to City (88% of passing accuracy).
For front players, it's just very difficult to perform if they know they'll get the ball only a couple of times in a game. For anyone who played football it's obvious how hard it is to get into a rhytm in that kind of game when you're having 30% possession, so I wouldn't blame forwards much for underperforming in that game.
Our record against top teams is abysmal this season, this might have insignificant impact on title race in the EPL, but we're not winning any Cup playing that way. We need to re-construct the midfield not only in terms of personnel but also in terms of strategy how to move the ball and get some control in the big games.
PS
Why do I think this is mainly a coaching isse? Because I see a players in good positions being ignored, because players who can do much more are just passing it backwards (and you can see this is the only thing on their mind), and because we seem the be playing very safe most of the time, while more often than not good things happen when we speed it up a little bit.
We only seem to play triangles with a few players like Bruno, Mata, van de Beek. Most of the time the passing is vertical or horizontal, just basic stuff.
Our rivals are less afraid of losing the ball than we do, and take more risks. We can do that as well, even with current personnel.
Last edited:
The "playing the same game 100 times" point is not something to take literally. It's just a way of explaining how probabilities work. Which is, the more times you replay a given scoring-chance, the bigger the chance that the result matches the probability. If you throw a dice 3 times, you may end up with all sixes. If you throw it a million times, you will end up getting all numbers a very similar number of times.
The reason why this matters is because it helps to differentiate in a more precise and objective way between what is luck and what isn't. Just like you can throw three sixes in a row, you can score three goals off three low-probability chances. 3 goals is 3 goals regardless, but in both cases, you've been lucky.
Our finishing is so bad, we were basically heading into a draw. The thing is, xG is a much stronger predictor of scoring than conversion rate. Meaning, the number of goals scored by football teams is generally closely aligned to the number and quality scoring chances they create. The ability of different teams to convert those chances varies much less. Like anything to do with probability, this only becomes clearly visible when you look at huge numbers of games, whereas pretty much anything can happen in one individual game. But that doesn't mean that those probabilities are any less in operation in any individual game than in others.
And, this is a really great thread by the way. Many thanks to Borys and other contributors!
Probably the first time I disagree with you here. Well, may be it's because both of our analysis aren't backed by stats but only purely based on what we watched.
First of all, I believe in first half was about 50:50, pretty sure Keane said similar thing that we played well in first half. I do think our chances are similar to their chances. For example, they had two offside goals (Gundogan & Foden), while we had Rashford offside (Stone's own goal) and Rashford good chance blocked by Dias. And KDB's hit the post is long shot which similar to Bruno's long shot. Others were City long shot and Maguire's header from set piece.
Second half was the big difference. I think our front four were underperforming compared to the first half while City maintained the same level of performance and since we conceded early goal from that set piece, I think we tried to play riskier by letting our double pivot to went up in advanced area which neglected by City solid defense (credit on them) and they used quick counter to punish us especially we don't have quick centre back to stop the counter.
In conclusion, there are four points I would like to make:
- First half was 50:50
- City maintained the same level of performance while we dropped ours in second half
- They hurt us lot on counter due to our double pivot tried to push higher and we are missing Bailly for this issue.
- City solid defense were crucial to nullified our front four and launched the counter
Last edited:
I think this the wrong way to interpret Total xG score.
It’s not if “we” played the match 100 times. It is if the match was played 100 times with random players who got the same chances to score. I think this distinction is important to make. When using xG, there is no correlation to the individual players involved, it is based on average performance in certain situations.
It doesn’t make it less valuable, however it doesn’t really say if a team was lucky or not.
Agreed and understood, but surely it does, at least vaguely, say if a team was "lucky" or not? If goals scored are very different from xG?
This is my view on it and other may disagree.
It might indicate if you're lucky or not in one game maybe, but if you watch these stats over a longer time I would say that variation evens out and the xG vs goals scored depend on other things. It could be an amazing midfield that serve the forwards with great passes, and the forwards make great runs but are poor at converting their chances in to goals. Then it is not poor luck, it could be a midfield that is much better than the forwards. The opposite can also be true. A team with clinical forwards know that they can play defensive and still win as long as they create 2-3 chances per match. Then it is not luck.
We saw one season how a world class goalkeeper could make a difference for United. We were expected to concede much more but DDG was amazing at stopping close range shots (exactly those shots that give high points in the stats).
Remember when people watched football to enjoy it?
Yeah like I said, I hate playing against Pep, it strongly affects my view as they looked a much better team, but I agree we had similar chances first half.
Clearly they have better players, but the whole team looks very good in the system they play. I'd say coaching is key for them right now.
They do look very solid defensively though - both as a formation, but also as individuals.
@A-man @justsomebloke
Quoting fbref definition on xG:
"Positive numbers suggest better luck or an above average ability (to score goals/stop shots)."
If you look at the np:G-xG at the end of the season, you will the best strikers topping that parameter (currently Son*, Calvert-Lewin, Bruno and Salah from forwards). It means they are scoring more goals then expected (considered the chances they had!). It is a better way to judge strikers - for example Greenwood and Martial were in top 3 last season, while Rashford oscillates around 0 meaning he's pretty average (scores goals because he gets many chances).
You can be lucky in one game, or in a good form for a while like Son is now (* 12 goals from 28 shots...). Achieving good xG for a few seasons is no luck, it's ability.
For us it means that if Martial and de Gea have underperformed massively this season, whether that's a good sign for the future or not is another matter. I'd say the former, as we're in a good place in the league!
xG is actually a good way at looking at freak games which happen every now and then (like Southampton scoring 2 goals with xG total of 0,4 or us winning against Brighton 3-2).
Yes the best forwards are in general better at converting chances in to goals. Then obviously the forward might be very good at positioning himself but poor at converting chances, then he will still get many goals. Luck is just normal variation and will even out in the long run. If a player is in good or bad form, that is something different.
Burnley 0 - 1 Manchester United (Pogba 71')
xG=1,2 xGA=0,9
Another very good game by Pogba, I was wrong about him and Matic in that game. Ole knew Burnley won't be playing a lot of football on the ground so it made sense to go with them as they did well in midfield. Seems like Paul has found some confidence and does not need to show off everytime he gets the ball, but instead does simple but effective things.
Miss of the Week Award goes to Anthony Martial, what puts him in front with a total of 3, followed closely by Rashford (2) and Bruno (1).
xG=1,2 xGA=0,9
- possession 62%
- Shots on target 7/13
Another very good game by Pogba, I was wrong about him and Matic in that game. Ole knew Burnley won't be playing a lot of football on the ground so it made sense to go with them as they did well in midfield. Seems like Paul has found some confidence and does not need to show off everytime he gets the ball, but instead does simple but effective things.
Miss of the Week Award goes to Anthony Martial, what puts him in front with a total of 3, followed closely by Rashford (2) and Bruno (1).
Liverpool 0-0 Manchester United
xG=0,9 xGA=1,5
We seem to have no clue how to move the ball through midfield. Or we simply don't want to, I'm not convinced if that's not the game plan. Anyway, I hope we work on keeping the ball because I think it's frustrating for the players we're not able to string a few passes together at times- especially in first 20/30 minutes.
To make the next step forward IMO Ole needs to stop tinkering with formation. It might be good for keeping all players fit, and this strange 4-2-----4 formation allows us to play both Pogba and Bruno together, but I guess at this point we need to consider if it's worth doing.
For example lets analyze for a while how we set up while in possession playing from the back:
Liverpool front three and two midfielders pressing high, which leaves a lot of space between back line and midfield. And this is what our front 4 does:
By the way, possible to draw wrong conclusions looking at the heatmap without context of possession: front 3 for both teams- looks like it's actually Liverpool forwards staying high, and United forwards dropping deep:
Overall I think description of Bruno performance sums up United in this game: rash and naive. Liverpool had twice as much possession and passes (693 to 365), which was probably caused not even by lack of ability to keep the ball, but rather unwillingness. It seemed like we needed to get the ball forward as quick as possible.
We start to look better defensively. Maybe not City-level yet, but I didn't have a heart attack every time Liverpool got close to the goal so that's a big improvement. At the same time we should be very happy sitting top with neither of front players performing well.
Still a lot of room for improvement for United, and we need to improve because we've been sliding lately, fine margins.
xG=0,9 xGA=1,5
- 34% possession
- shots on target 4/8
- Fernandes has been pretty poor lately, I don't worry much and don't blame him but he's had a terrible game- 57% passes completed (only 53% in short passing) ; no dribbles won; did nothing defensively except for 1 clearance; dribbled past twice. Stats don't show how many wrong choices he made. The problem with him is that he's more dangerous for us (losing the ball in dangerous positions) than for opponent. It's basically the Pogba case from beggining of the season.
- contrary to my opinion after the game, Pogba was poor as well (offensively). 51% passing accuracy; only 1/10 accurate long pass; dispossessed 3x (top) and 5 miscontrols (top). He was decent defensively though: 3 interceptions, 1 tackle. I am not sure if it makes sense to play both him and Bruno at the same time.
- for a player who is considered a DM (not sure I'd agree with that label), Fred is quite easy to go past: was dribbled 7x (top, McTominay second). Still had by far the most tackles and joined most interceptions (with Pogba). To be honest I didn't see him struggling much, I thought he had a good game apart from a few technical errors (considered as misplaced passes I guess).
- McTominay had a good game, his moves helped us to regain some control after first 30 minutes when we couldn't get hold of the ball (as expected). Won 3/3 aerials; was the only player from midfield who was on "normal" level of passing accuracy (82%); second highest succesful pressings (42%, after Shaw). He is a good passer, his problem is he takes too long to make a decision. It's not the lack of accuracy that kills him.
- Fantastic game from Shaw. Looks by far the most in-form player right now. He seems to be our "safety switch" when we're in trouble and don't know what to do with the ball (which is often): most touches in the team, most passes, decent accuracy (77%, 95% of short), won all 3 headers, most recoveries.
We seem to have no clue how to move the ball through midfield. Or we simply don't want to, I'm not convinced if that's not the game plan. Anyway, I hope we work on keeping the ball because I think it's frustrating for the players we're not able to string a few passes together at times- especially in first 20/30 minutes.
To make the next step forward IMO Ole needs to stop tinkering with formation. It might be good for keeping all players fit, and this strange 4-2-----4 formation allows us to play both Pogba and Bruno together, but I guess at this point we need to consider if it's worth doing.
For example lets analyze for a while how we set up while in possession playing from the back:
Liverpool front three and two midfielders pressing high, which leaves a lot of space between back line and midfield. And this is what our front 4 does:
By the way, possible to draw wrong conclusions looking at the heatmap without context of possession: front 3 for both teams- looks like it's actually Liverpool forwards staying high, and United forwards dropping deep:
Overall I think description of Bruno performance sums up United in this game: rash and naive. Liverpool had twice as much possession and passes (693 to 365), which was probably caused not even by lack of ability to keep the ball, but rather unwillingness. It seemed like we needed to get the ball forward as quick as possible.
We start to look better defensively. Maybe not City-level yet, but I didn't have a heart attack every time Liverpool got close to the goal so that's a big improvement. At the same time we should be very happy sitting top with neither of front players performing well.
Still a lot of room for improvement for United, and we need to improve because we've been sliding lately, fine margins.