The Premier League got back into action on Boxing Day following a six-week break, and it certainly did not disappoint. With every team now having played at least 15 games, it is fair to be able to draw some conclusions on which teams are so far either performing as expected, underperforming or overperforming in relation to expected goals scored and conceded. What indiations may be drawn? Find out below.
Note: Throughout this analysis, own goals are always omitted and so to penalties unless indicated otherwise. The source used for expected goals (xG) figures is FBRef.com, but it is important to beware of the fact that different websites use different models to calculate xG so going on different sites might give you some variations with the figures used here. xG indicates the probability of a particular shot resulting in a goal, so an xG figure of for example 0.6 indicates a 60% probability of that shot being scored. Non-penalty expected goals (npxG) relates to expected goals excluding any penalties awarded or given away.
Expected goals scored (xG) excluding penalties: which teams have the best figures and what is the correlation between this table and actual league position?
In terms of non-penalty expected goals scored (npxG), Manchester City currently boast the highest figure, with a total of 29.3, which equates to 1.95 per game. Arsenal, the current league leaders, have the next best figure at 27.8. This correlates well with the fact that the better performing teams create a considerable number of chances, and good quality chances that should probably result in goals. In fact, the top 7 teams for npxG (both as an absolute total and per match) are all currently the top 7 in the actual Premier League table.
Overall, the correlation between the table of npxG and the actual league table is reasonable, with more than half of the teams (in fact 11) only having a difference of 1 in either direction (so either 1 position higher up or lower down) between their position in the npxG table compared to the actual, suggesting that npxG is a decent measure of where a team stands. In fact, Wolverhampton Wanderers, Leicester City and Brighton & Hove Albion’s position in the npxG table is the same as their current league position (18th, 13th and 7th respectively).
That is not to say that there are no outliers in this respect, however. A notable example of this is Chelsea, whose npxG figure ranks only 15th, while their actual league position is 8th. The most extreme positive case (meaning that their actual league position is better than that in the npxG) is Crystal Palace, who are 19th in the npxG but 11th in the league, a difference of 8 positions. On the other hand, the most extreme negative case is Leeds United, whose npxG position is 8th but their actual league position is only 15th, 7 places worse off. Another notable struggler is Everton, who are currently 5 places worse off in the league than their rank in the npxG table.
What do these outliers suggest purely by looking at league table position compared to npxG table? One suggestion is that the overachievers (i.e., those teams with an actual league position better than their rank in the npxG) could be very strong defensively so do not need to score that many goals to get positive results in the first place. Another one is that they are simply overachieving and have been somewhat “fortunate” that other teams are underperforming, be it against such teams or even overall and this would indicate cause for concern for the rest of the season if opponents were to be more ruthless with their chances.
Actual goals scored (excluding penalties) compared to npxG: which teams score more than they should, and which teams simply do not score as much as they should – and what does this imply? Does this possibly explain why some teams are higher up or lower down in the table compared to the npxG rank?
In this respect, the team with the best absolute difference between goals scored and npxG is Manchester City, with a difference of 9.7, having scored 39 goals excluding penalties given a figure of 29.3 npxG. This indicates that in fact, Manchester City are remarkably good at finishing chances that might not necessarily be of the highest quality, with Erling Haaland being the perfect demonstration of this point, having scored 17 goals (excluding penalties, where he scored 3) with a total of 10.6 npxG (the most of any player in the division by a fair distance), a difference of 6.4 which is the biggest of any individual player, and accounts for 66% of Manchester City’s difference between npxG and actual non-penalty goals scored.
Arsenal’s difference between total npxG and actual goals scored is the next best at 6.2, suggesting that they are extremely efficient at scoring as well. However, an interesting feature here is that Gabriel Jesus has a total npxG of 7.7 (only Haaland has a higher figure) yet has only scored 5 goals indicating a level of wastefulness on his end – while on the other hand, Gabriel Martinelli’s total npxG is just 3.1 yet he has scored 6 goals. In fact, all the current top 4 rank within the top 6 for the highest difference between goals scored and npxG, which highlights how elite players, and the best teams, require few chances of high quality to score.
However, in percentage terms, the team with the best difference between actual goals scored and npxG is AFC Bournemouth, whose figure for goals scored is 148.8% (as shown in the second to last column) of their npxG (and in fact, their difference between actual league position and npxG rank is the second highest at 6). This is explained by the fact that they have the lowest npxG in the division, at just 12.1, yet have still managed to score 18 goals – mostly thanks to Kieffer Moore and Philip Billing, their top scorers so far this season with 4 who have scored a total of 8 goals combined with their total npxG being just 3.3 between them. Also, this highlights that AFC Bournemouth need to create more and better-quality chances to increase their likelihood of scoring which would in turn lead to increased probability of winning games and staying up, as they simply cannot indefinitely rely on Moore and Billing drastically exceeding their npxG total.
In the case of Crystal Palace, their npxG is the second lowest at just 12.8, but they have managed to score 14 goals, highlighting a level of efficiency from their end at scoring despite minimal creation of good chances. They have been awarded 2 penalties this season, yet failed to convert either of them (granted, they did score on the rebound from one of them but for statistical purposes it is not considered a converted penalty), so they have in fact scored a total of 14 goals from an overall xG of 14.4, which is almost exactly in line but highlights that there would have been a larger discrepancy had they converted either (or both) of their penalties.
Another interesting case of a team who has scored more goals than their npxG would suggest is Liverpool. Earlier in the season, Liverpool won 9-0 at home to AFC Bournemouth, in a game where their npxG (which in Liverpool’s case is equal to overall xG due to having not been awarded a single penalty) was just 3.3, meaning they outscored their npxG by 5.7 goals, indicating that in fact, had they scored approximately the same level of goals as npxG in that game (e.g., 4), that would have led to their goal total being considerably fewer than npxG, highlighting the freak nature of that win which has almost artificially boosted their goals scored figure and also a possible level of wastefulness that might not be so obvious on the surface.
On the other hand, the team with the biggest absolute difference between npxG and actual goals scored in negative terms is West Ham United, who have a npxG of 16.4 yet have only scored 10 goals excluding penalties, therefore a difference of -6.4, indicating that West Ham have been extremely wasteful with their chances and been let down by poor finishing (possibly mitigated also by some extraordinary goalkeeping).
In fact, this figure would be even worse for them if penalties had to be included, as West Ham United have been awarded the most penalties this season (5 together with Fulham), yet converted only 3, which is a poor return given that the xG for a penalty is roughly 0.78 (in turn suggesting that West Ham should have scored at least 4 out of their 5 penalties). The player contributing most to this problem has been Jarrod Bowen, who has a figure of 3.2 for npxG yet has only scored once – although Tomas Soucek and Lucas Paqueta having scored 1 and 0 goals from a total of 2.5 and 1.3 npxG respectively.
Other teams with an extreme negative difference between non-penalty goals scored and npxG include Everton, Wolverhampton Wanderers and Nottingham Forest, with an absolute value total difference of 6.1, 6.1 and 5.7 respectively. This indicates that all these teams are extremely wasteful with their chances created, and even in percentage terms these 4 teams have the lowest percentage of goals relative to npxG, with Wolves being the worst at 56.7% (having scored just 8 non-penalty goals from a total of 14.1 npxG, which is a low npxG total anyway as only 2 teams have a worse figure), and in fact no team has scored fewer goals than Wolves in the division. Also, Wolves’ percentage of overall goals relative to xG is boosted significantly by the fact that they have been awarded 2 penalties and converted both (thanks to Ruben Neves).
Five of the six teams with the worst difference between non-penalty goals scored and npxG are in fact the bottom 5 teams (Southampton having the 6th worst difference between non-penalty goals and npxG in absolute value and percentage terms, although it is considerably better than the other sides), proving that it is little wonder why these teams are struggling in the league, despite in some cases having relatively good xG and npxG.
Expected goals conceded (xGA) excluding penalties: which teams have the best figures and what is the correlation between this table and actual league position?
In relation non-penalty expected goals conceded (npxGA), Manchester City currently have the best figure, with a total of just 10.2 and just 0.68 per game. Arsenal’s figure for npxGA is the next best, standing at 10.5 in total and 0.7 per game. Brighton & Hove Albion have the next best total figure at 13.7, but although Newcastle United’s figure is slightly higher (14.1), they have played one more game compared to Brighton and so their average npxGA per game is lower (0.88 for Newcastle compared to 0.91 for Brighton). This therefore correlates well with the fact that the better teams concede less, as they likely have better defences and so are less likely to give away quality chances which could then result in a goal. In fact, 7 of the best 8 teams in this metric (therefore with the lowest npxG) are the top 7 in the league currently.
That said, in this respect, there are more outliers and in fact only 5 teams have only a difference of 1 in either direction (so either 1 position higher up or lower down) between their position in the npxGA table compared to the actual. Therefore, this suggests that the correlation between npxGA and league position is not as strong, especially compared to npxG.
The most extreme cases here in either direction concern West Ham United and Fulham, with the former ranking 5th in the npxGA but 16th in the league table, a difference of 11 positions negatively, with Fulham on the other hand having a difference of 11 positions positively (ranking 20th in the npxGA table but 9th in the league table). In West Ham’s case, this is even more extreme when one considers that in total, they’ve conceded a total of just 20 goals (including 1 own goal and 2 penalties), which is joint 7th best, but one can explain it by their wastefulness in terms of scoring highlighted earlier (having scored considerably fewer non-penalty goals than npxG). Fulham’s figure is mainly explained by their ability to score several goals (in fact their overall xG position correlates very well with their league position), which can help them get away with conceding a higher number of goals.
Actual goals conceded (excluding penalties) compared to npxGA: which teams concede less than they should, and which more? What are the implications? Does this possibly explain why some teams are higher up or lower down in the table compared to the npxGA rank?
In this respect, it should firstly be pointed out that Newcastle United have conceded the fewest goals of anyone in the division, having suffered just 11 so far (with 1 of those being a penalty). Furthermore, they have in fact conceded 4.1 goals fewer than their npxGA total of 14.1, having conceded 10 goals (excluding the penalty) in all, which is the fourth best in the division. In their case, this indicates two things: firstly, the fact that they are solid defensively due to a low npxGA in the first place, but also that opposition have been poor at converting their chances (or that they have on occasion benefitted from good goalkeeping by Pope).
Only 2 teams have had a more favourable difference between npxGA and non-penalty goals conceded than Newcastle, one of whom is Everton – who have conceded just 17 goals from a total of 24.7 npxGA (a difference of 7.7). In this case, given that only Fulham have a higher figure for npxGA than Everton, this arguably indicates more that Everton have been rather fortunate that the opposition have been wasteful with chances and/or that they have had Pickford in goal bailing them out on occasion (with Everton having a save percentage of 80% which is only bettered by Newcastle who have faced fewer shots in the first place), rather than any indications of defensive solidity. In fact, Everton have in total conceded 19 goals (2 of which were penalties), with only 5 teams conceding fewer – and on the surface that looks solid – but when looking at npxGA, one can clearly see that Everton have been somewhat fortunate to not have conceded more, and in percentage terms, Everton’s total goals conceded represents 68.8% of their npxGA, the best of all this season.
The other team with a more favourable difference than Newcastle between npxGA and non-penalty goals conceded besides Everton is Chelsea, who have conceded just 15 non-penalty goals despite having a total of 19.9 npxGA. This is somewhat concerning for Chelsea, as their npxGA figure ranks just 12th in the league, with even Wolverhampton Wanderers, Everton and West Ham United (all of whom are in the bottom 5) having a superior figure, which could possibly indicate that Chelsea have been somewhat fortunate to not be further down the table given that they have in effect, conceded fewer goals than they should have (and on the flip side, Chelsea’s total non-penalty goals slightly exceeds their npxG figure).
On the other hand, the team with the least favourable difference between their actual goals conceded and npxGA is Nottingham Forest, with totals of 31 (excluding penalties) and 22.8 respectively, a difference of 8.2. This also further goes to show why Nottingham Forest are evidently struggling so far this season, as besides being extremely wasteful with their chances going forward (highlighted earlier by the fact that their total non-penalty goals scored is 5.5 lower than npxG with only 3 teams having a worse difference), they are conceding considerably more goals than they should, with only Everton and Fulham having a worse npxG figure than them. Reasons for this include a likely combination of individual errors (either defensively or goalkeeping), and outstanding finishing by the opposition.
The team with the next least favourable difference (and the least favourable overall in terms of percentage, slightly ahead of Nottingham Forest) between their actual goals conceded and npxGA is Southampton, who have conceded 28 goals despite only having a total of 20.5 npxGA (a figure which is better than 7 other teams). This is arguably the biggest reason for Southampton struggling so badly this season (in fact they are bottom of the table), as although their npxG is not the highest and their actual goals scored is only slightly lower than npxG total (it’s a higher figure in percentage figures, though), they have conceded far too many goals relative to their npxGA, clearly hinting at a combination of errors on their part and quality finishing by opposition, in a similar vein to Nottingham Forest.
AFC Bournemouth are another team who have an unfavourable difference between npxGA and actual goals conceded, having conceded 27 non-penalty goals from a total of 22.1 npxGA, although this can very evidently be explained by their 9-0 defeat to Liverpool (mentioned earlier) where they had a total of 3.3 npxGA but ended up conceding 9 goals. In fact, had they conceded roughly as many goals as expected (e.g., 4), their total non-penalty goals conceded would be just 22 from a total of 22.1 npxGA, far more in line with expectations.
Interestingly, despite having the lowest npxGA of anybody at 10.2, Manchester City have conceded 12 non-penalty goals, an unfavourable difference of 1.8 (and non-penalty goals conceded represents 117.6% of their total npxG, with only 6 teams ending up worse off in percentage terms). This can mainly be explained by the fact that Manchester City do not suffer shots with a high npxGA in the first place, and so any goal conceded has a stronger possibility of exceeding that total.