The Stats That Show The Trouble Leeds United Find Themselves In
In this piece, Josh Hobbs explores the stats behind Leeds United’s season so far, breaking down the numbers behind their poor start…
Leeds United lost 1–0 against Southampton on Saturday, meaning that the Saints leapfrogged them, leaving the Whites in 17th place and just 3 points above the relegation places by the time the game-week was over.
As many have pointed out, Marcelo Bielsa’s side was shorn of its star players as Raphinha, Kalvin Phillips and Patrick Bamford were all absent. Additionally, Luke Ayling, Robin Koch and Junior Firpo were also out due to injury. All of those players, possibly apart from Koch, would have started the game if they’d been fit, and nobody would argue that their absences didn’t affect how Leeds played.
However, the performance delivered was an appalling one and not one that Bielsa himself wanted to blame on absences.
I don’t want to dwell on that performance now, but I do want to explore a little of Leeds’ current strife. I don’t see it as being as simple as the team finding form when those players return. There are issues here, and there have been for many weeks now.
To illustrate these problems, here is how data forecasters fivethirtyeight predicted the Premier League table to finish before a ball was kicked this season:
Using their team rating system, which awards each team an attack and defence score (calculated using expected goals as well as real goals scored and conceded), as well as an overall ‘SPI’ rating, fivethirtyeight runs 1000s of simulations of the season, giving an average simulated season result and probabilities for relegation, Champions League Qualification and a title win.
As you can see, Leeds were given an overall rating of 72.6, with an attack rating of 2.0 and a defence rating of 0.8 (in the latter case, the lower the number the better). This led to a predicted finishing position of 11th and only a 10% chance of relegation, just 2% more likely than qualification for the Champions League.
This represented a drop of 9 points and 2 places below where the Whites finished last season, but if this is what was offered before the season — knowing how important consolidation in the league is to a newly promoted side — one would imagine that the vast majority of fans would have jumped at it.
This is how fivethirtyeight’s most recent sims look after the Southampton loss:
Both the attack and the defence ratings are now trending the wrong way, with the attack rating dropping to 1.9, whilst the defence rating is now 0.9.
The predicted finishing position has dropped 5 places to 16th, with a points total 8 fewer than originally predicted. Perhaps most worryingly, the relegation probability has now doubled to 20%.
If you don’t think this season is turning into a relegation scrap, it might be time to think again, unfortunately.
Expected Goals For and Against
So, what’s gone wrong this season? Looking at expected goals for and against might be helpful here, as it can give us a sense of how Leeds have performed in terms of chance creation and prevention, instead of the more noisy statistic of goals for and against, which is influenced a little more by luck.
Firstly, let’s look at how Leeds performed for non-penalty expected goals for and against last season, where they performed well enough to finish 9th in the table.
Here is non-penalty expected goals for:
As you can see, Leeds were one of the best in the whole division for this metric. To see a newly promoted side performing better than the side which finished 2nd in the league (Manchester United) for NPxG is frankly ridiculous.
Leeds ultimately scored 60 goals last season — 4 more than Chelsea and enough to make them the league’s sixth highest goal scorers.
Now looking at NPxG against:
As you can see, this is a lot less pretty as Leeds were the third-worst in the league when it came to preventing chances on goal.
However, their excellent attack meant that they could make up for their poor defence and they were more than worth a mid-table position, according to expected goal difference per 90. Note — this graph does not discount penalties.
Although it’s slightly unfair to compare 8 games to 38 games, here are the same charts from this season to give you an idea of the direction Leeds’ underlying numbers have been going in.
Again, here’s NPxG first:
Leeds are now 8th best in the league for this metric, creating 1.3 NPxG in 21/22, compared to 1.43 in 20/21.
Unfortunately, it gets worse when we look at the defensive numbers.
This season, Leeds are only just better than Norwich when it comes to NPxGA, putting them 2nd bottom of the league. According to the metric, Leeds are giving up chances worth the best part of 2 goals per game (1.74 NPxGA compared to 1.47 in 20/21).
As you can see from the xGD graph next, their expected goals for value isn’t dragging them up into such a strong position this time around.
According to this graph, Leeds rank as 5th from bottom, just one place above where they currently sit in the table (-0.53 xGD compared to -0.14 in 20/21). Notably Arsenal are just below, but whilst everybody can see that Arsenal are performing poorly, it’s clear that Arsenal’s squad depth and quality of players can help them perform above the level of relegation strugglers, even when their underlying numbers are very poor.
Clearly, we can’t say the same about Leeds right now. The Whites are pretty much in the position in the table that their expected goals difference suggests they should be.
That shows us that something fundamental is broken this season. In the most simple terms, the team is not creating as many good chances as they did last season and they’re giving away more. There’s no way to dress that up as not being concerning.
Attacking Metrics Down Across The Board
Unfortunately, it’s not just expected goals which is down this season. Pretty much every attacking metric across the board is worse than 20/21.
For the sake of not making this article more depressing than it is already, I won’t go into all of them but I’ll mention a few that I think are important.
Notably, shots per 90 are down this season, but only by 0.2, as last season was 16.8 and this season it’s 16.6. However, Leeds are taking their shots an average of 2 yards further away this season, as their average shot distance is now 18.4 — the 2nd furthest in the league after Norwich — compared to last season’s 16.2, which was the second shortest distance in the league.
This demonstrates how Leeds are struggling to get into the box to create the types of chances they have in the past. We are very rarely seeing chances from cutbacks to the penalty spot or balls across the six-yard-box begging to be finished.
Instead, we’re seeing many long-shots fly into the stands. As evidence of this, just go and watch all the goals Leeds have scored so far this season and you won’t find many which are ‘repeatable’. By which I mean, they don’t come from clear attacking moves and passing patterns, which differs hugely from previous Bielsa seasons.
This leads me to my next point: shots on target percentage. I believe it’s partly to do with the previous issue of shots being taken too far away from goal, but it’s a fact that shots from outside the box aren’t hitting the target as often. They miss the target completely, or are blocked before reaching goal far more of the time. Last season Leeds averaged 37.4% of their shots hitting the target. This season it’s down at 27.5%. This isn’t good enough and these issues have led to Leeds scoring far less often.
Last season, Leeds just over-performed their NPxG per 90, scoring 1.47 compared to the 1.43 underlying number. However, this season Leeds are considerably below par, scoring 0.88 per 90 compared to an NPxG of 1.3. Due to how bad the defensive numbers have been throughout Leeds’ time in the Premier League, this issue of the attack breaking down is the key problem, as Leeds simply can’t outshoot their poor defence anymore.
So, why can’t Leeds get into better positions to create anymore?
Bielsa’s style has always relied on getting the ball quickly from back to front and Leeds made hay last season with fast attacking transitions. This season however, the team seem to be rushing many of these transitions, and are often losing the ball before they manage to actually get it into a position to create a dangerous chance.
Recently, I wrote about Raphinha forcing it a little too much as he’s been playing many long balls from deeper positions, when I want to see him high up the pitch, doing the kind of damage we saw him do for Brazil in the international break.
Looking at possession stats from this season, I found that Leeds are averaging fewer passes attempted this season, but the percentage of which that are long passes has increased by 3%. This isn’t a huge amount but it certainly felt worth mentioning.
The big loss in Leeds’ progression has been ball-carrying though. Last season, The Whites achieved an average of 43.4 progressive carries per 90. These are defined as carries that move the ball at least 10 metres closer to the opponent’s goal. In 21/22, that figure stands at just 35.9. This is a big drop off.
Last season, Leeds’ chief ball progressor through carries was Luke Ayling. In 21/22, he seemed to start the season in poor form, then covered at centre back due to injuries to Diego Llorente and Robin Koch, before he picked up an injury himself. As such, it’s not worth looking at his stats for this season as the sample size will be too small, but it’s clear that Leeds have missed his carries up the right flank, as much as I like Jamie Shackleton.
I would suggest that because these carries haven’t been there, this has pulled Raphinha into a deeper position, taking him away from the position Leeds really want him to be — isolated against a full-back so he can do what he did to Charlie Taylor against Burnley.
Hopefully Ayling’s eventual return from his minor knee surgery will see him come back refreshed and he will assist in improving the ball progression. However, the concern is where Leeds might be by the time he returns.
I won’t go into the defensive numbers properly as whilst they are a bit worse, they were basically terrible last season as well.
Even in the run of games at the end of the season when Leeds didn’t concede many goals, the underlying numbers were often bad. A combination of Meslier’s saves and bad finishing meant that Leeds conceded just the 7th most goals against, despite giving up the 3rd highest amount of expected goals against.
This season, I would suggest that Leeds are being punished more because they are losing the ball in transition far too often. They are then being hit on counter-attacks with the man-marking system all over the place. Armando Broja’s winner on Saturday, Michail Antonio’s winner for West Ham, and Allan Saint-Maximin’s equaliser in the Newcastle game are stark examples.
As mentioned though, Leeds have previously covered for their bad defensive numbers with a great attack. Now they have an underperforming attack, they can’t get away with the bad defence anymore.
So How Do They Turn It Around?
It’s difficult to say for sure. As I said at the start, I don’t think it’s as simple as ‘players coming back will make everything fine’, as I’ve demonstrated that the numbers have been bad all season. Whilst there have been injuries throughout, the rest of the games haven’t seen the squad as decimated as the Southampton game and Leeds have still only won once.
I mentioned in the article I wrote before the transfer window closed that Leeds would likely see a downturn in performance from players who have played huge amounts of minutes over several seasons, and I think we’ve seen that from Stuart Dallas and Luke Ayling. Leeds missed their opportunity to bring in a midfield reinforcement which was desperately needed and the squad will now have to improve.
Of course, there is something the individuals would bring to the table. Bamford was by far Leeds’ top contributor when it came to xG last season and he has missed the last 3 games with injury. Rodrigo has not been the same presence in the box and the return of the number nine would likely see Leeds become a greater attacking threat.
Without him in the team, it’s hard to imagine anybody else scoring the type of goal that Bamford did to allow Leeds to pick up a point at Turf Moor. With him there, there’s a target for balls into the box that just isn’t present when he’s not playing.
Likewise with Raphinha — it’s hard to imagine Leeds creating so little against Southampton with him on the pitch. He’d likely have conjured at least one opportunity from somewhere, as he did against Burnley when nothing looked like happening.
However, it’s concerning that Leeds are totally reliant on one player for their goalscoring and one player for their chance creation. That’s not a system that’s working well.
It should be enough to see Leeds finish above Norwich, Watford, Burnley and Southampton though, as none of those teams have good attacking systems, nor the individual quality that Raphinha and Bamford possess. When you add in Phillips and Meslier as well, those are 4 players that the true bottom teams don’t have anybody near the quality of.
As long as none of those players pick up bad injuries, one would expect Leeds will eventually begin to climb away from the bottom 3, but don’t count on anywhere near the level of finish from last year. It’s looking more like 14th is an absolute best case scenario with 15/16th a lot more likely.
If that ends up being the final position, the club will chalk this season up as a consolidation season, but some serious work will be needed in the summer — particularly in the midfield — to avoid a relegation fight next season, let alone think of pushing up the table.