Assessing the Championship Top Four

Over the course of the 19/20 Championship season, I have been tracking Leeds United’s goal difference and how it correlates to their expected goal (xG) difference as a way of attempting to show Leeds’ consistency in their underlying numbers, despite a couple of rocky patches in terms of results.

On Sunday afternoon, I shared the graph on the All Stats Aren’t We twitter account and wrote a short thread about it. Quite a few of the replies that came in wanted to see how Leeds’ closest rivals would look on similar graphs, so I decided to give the people what they want (with the help of one of our followers gathering data) and I will attempt to assess the Championship’s automatic promotion contenders going into the final ten games of the season.

Just a few weeks ago, I personally felt that Brentford would be the biggest threat to the Top Two. However, since their draw against Leeds at Griffin Park their form has fallen off the edge of a cliff: only 3 points picked up in their last four games, leaving them 11 points behind Leeds with ten to play.

Anything can happen in football, but the likelihood of them making up that gap at this stage is tiny, so I’ve decided to limit this piece to looking at West Bromwich Albion, Leeds, Fulham and Nottingham Forest.

West Bromwich Albion

First up are the league leaders, West Bromwich Albion. After a run of one win in eight games beginning in the Christmas period, they turned it around at the beginning of February and won five of the next seven.

The question to ask is whether those two runs of form are reflected in their underlying numbers? We turn to the goal difference vs expected goal difference graph in order to find out.

As you can see here, after a slow start to the season — they only picked up three wins out of six games in the first month — they began a period of consistently over-performing what were some fairly average numbers in terms of expected goal difference, turning games that they should have dropped points in according to chances created into points. In fact, their results were so good in the early part of the season that they didn’t lose until October (1–0 to Leeds at Elland Road).

They then went unbeaten for their next 15 Championship games, which is initially reflected in a rise in their expected goal difference but quite quickly drops off, which showed that it was an unsustainable run of form. They were winning games whilst their opponents were creating better chances than them and eventually it was bound to bite them.

It was at that point that they embarked on their aforementioned run of one win in eight games. Interestingly though, their performance in terms of expected goal difference turned around earlier than their results. On the graph, we see a big drop in their goal difference as their expected goal difference began to climb again. This was due to a run of three games where they won the xG battle clearly, despite losing two and drawing the other.

Their most recent good run of form was also their most consistent in terms of underlying numbers; rather than winning games they perhaps didn’t deserve to, they were clearly the better side in all of them as new signing Callum Robinson, on loan from Sheffield United, gave them a new element to their attack. However, they will note with a bit of concern that their most recent game saw Wigan come to the Hawthorns and take the three points, not in a smash and grab, but by thoroughly outplaying and outshooting the league leaders.

How are they overperforming this much?

West Brom’s season has been defined by their over performance of their expected goal difference. They have scored nine more goals than expected and conceded three fewer than expected. How have they managed that over the course of the season?

There are three factors here:

Firstly, they have been clinical. They have scored 11 goals from outside the penalty area, which puts them joint top of the league with Derby for that metric. Shots from outside the box are valued much lower according to xG as they aren’t often scored.

However, West Brom are able to call on some of the league’s best forwards, like Matheus Pereira and Grady Diangana. They have also had regular goal scorers from all over the team. Their top scorer, Hal Robson-Kanu has netted only ten times, but they have five other players who have scored five or more.

Secondly, Sam Johnstone has over-performed in goal. Wyscout have a goalkeeper’s metric called ‘expected conceded goals’ or xGC which only accounts for shots on target faced. Johnstone has conceded 37 from an xGC of 39.04. Whilst this isn’t a remarkable over-performance, that difference of two goals is crucial considering that West Brom’s defence has not been particularly tight for a Championship title-winning contender.

Finally, they’ve been lucky. Many will argue that teams make their own luck in football, but it’s widely accepted in statistical analysis that luck plays a big part in football. Considering the number of variables in a fluid sport like football, it’s impossible to deny.

In the case of West Brom, they’ve been lucky that their opponents have done poorly with the chances they’ve been given, whilst being fortunate that they’ve scored some of the low probability chances they’ve had. As I’ve mentioned above, the quality of players at their disposal definitely makes a difference, but for it consistently happen throughout the season is still lucky.

In the period of time where their xGD is consistently going down but the GD is consistently going up, West Brom were extremely lucky that their good form lasted so long with such underlying performance issues.

Leeds

In the case of Leeds, the story is very different as they have been chronic under-performers of their expected goal difference.

As you can see, Leeds struggled to keep up with their expected goal difference from a very early stage of the season and it has consistently and rapidly diverged from actual goals. Note that West Brom’s expected goal difference peaks at +18.85, whilst Leeds hit the heights of +35.83. Leeds are underachieving their xGD by 11.83, though, whilst West Brom are beating theirs by 9.45.

All of this only adds up to a one-point advantage for West Brom in the table though. The league leaders will need to continue with their over-performance in order to stay on top, unless Leeds should drop away in their xGD performance again which is what we can see happening on the graph at the point Leeds’ GD line takes a dive.

As you can see, their expected goal difference still continued to climb during Leeds’ terrible run of two wins in ten games (when the goal difference line tumbles). As long as Leeds’ underlying numbers continued to be good, it was always likely that they’d turn it around in the end, which they have done, going five unbeaten and winning their last four without conceding a goal.

Is Leeds’ awful finishing the only reason for such disparity in the two lines?

Leeds are the best team in the league in terms of xG created (66.62) and have the lowest xGA in the league (28.94) so it is no surprise to see their xGD consistently climbing. They really should be out of sight in the league, not a point behind West Brom and sitting in second. The question to ask is why are they underperforming by such a large margin?

If you’ve watched the Championship at all this season then you know Leeds’ problems in front of goal. According to Wyscout, they’ve scored almost 13 goals fewer than expected. To save going into too much detail on this, I explored this issue a few weeks ago and you can read that here.

However, despite underachieving their xG to a huge extent, after a little wobble in September and October, Leeds went on a seven-game winning streak: represented by their goal difference line increasing sharply on the graph. The end of this run brought to a head the second reason that they find themselves so far below their expected goal difference: individual errors.

During Leeds’ horror run of two wins in ten games, Kiko Casilla conceded 20 goals from only 13.52xGC. Whereas earlier in the season Leeds had been covering for their poor finishing by shutting out their opponents, Casilla letting almost every shot he faced go past him meant that Leeds’ problems in front of goal were coming back to bite them.

It should be noted that Casilla had seemed to come out of this poor patch, keeping three clean sheets from 1.27xGC. However, with him now banned for eight games after being found guilty of racist abuse by the FA, Leeds now hope that youngster Ilan Meslier can be a safe pair of hands until the end of the season.

Fulham

In third place, sat five points behind Leeds, are Fulham. They’ve been fairly inconsistent throughout the season, with a run of dropped points never far away.

At the end of Leeds’ seven-game winning streak, Fulham were ten points behind Leeds and had Leeds not thrown away a 3–0 lead against Cardiff on that day would have found themselves 12 points back. As it happened, Fulham then put together their best run of the season just as Leeds went on their worst, closing the gap to just goal difference before going on another poor run, taking only two points from three games and allowing Leeds to open up a five-point lead.

Looking at their performance in comparison to their expected goal difference, Fulham have wildly over-performed, helped by a clinical 4–0 win over Millwall in the first month of the season when the xG suggested a 2–1 win would have been a fairer result.

After that, the lines ran in parallel with each other for a few weeks before, midway through the season, they began to pull away again as they beat Derby 3–0 in a game the xG suggested might have been 2–1 and as they came out 2–1 winners against Swansea despite the xG having it down as a draw.

In recent months, they’ve managed to win against Hull despite only taking 2 shots from in the penalty area, beat Huddersfield 3–2 in a game xG had as a 2–1 win for Huddersfield and won 2–0 against Preston despite xG calling the game even.

A quick glance at the graph would make Fulham’s goal difference look very impressive in comparison to Leeds’ and similar to West Brom’s but it’s important to check the scale. Fulham peak at +14 whereas West Brom peak at +28 and Leeds hit +24, despite massively underperforming.

This is all Mitro, right?

When it comes to Fulham’s over-performance, you’d be forgiven for presuming it’s because they’re clinical in front of goal.

In fact, they are actually 3.38 behind their xG for the season and it’s in their xGA that they have exceeded expectations. With 37 goals conceded, they have conceded 6.7 goals fewer than their 43.7 xGA. Marek Rodak has been key here, conceding only 22 goals from 28.03 xGC after he came in to replace Marcus Betinelli, who had conceded 15 goals from 12.15 xGC.

A note on their performance in front of goal — they have the top scorer in the league in Mitrovic, who is over performing his xG by twi, but he has 45.09% of Fulham’s goals. They have had only 8 goal scorers all season so they need others to carry the load should he hit a poor patch before the end of the season.

There can be a feeling that it’s Mitrovic or bust for Fulham. For reference, Hal Robson-Kanu has scored 15.63 % of West Brom’s goals and Patrick Bamford has 22.22% of Leeds’. They could do with somebody stepping up to carry some of that load.

Nottingham Forest

The last team we’ll look closely at are Nottingham Forest, who sit in fourth place. They have had perhaps the strangest season of all as they didn’t lose until October but have only put together a run of three wins in a row once in the whole season with draws coming much too regularly.

As you can see, their expected goal difference has been poor all season but they have somehow managed to beat it by a large margin. Infogol place them in an expected position of 21st in the league. With an expected goal difference of only +1 it is remarkable that they find themselves in the conversation of Top Two contenders with ten games left to go in the season.

As I mentioned at the start of this section, Forest have drawn too many games to be closely on Leeds’ tail. However, in many of those games, they lost the xG battle but came away with a point, which has kept their goal difference from dropping towards their expected goal difference. More importantly for them, it kept them ticking over in terms of points accrued.

It’s notable that around the half-way stage in the season, their xGD really dropped off and which led to them going on a run of one win in seven games, with that being a 4–0 win at QPR where they were particularly clinical. It was after that point that they went on their only three-game winning streak of the season and won five of their next sevengames. This brought them into the promotion picture as West Brom and Leeds were fading at the same time. Again though, they have struggled to string wins together and have only won once in their last five games.

Brice is right

Hopefully, you will have noticed that Forest’s expected goal difference has hovered around 0 throughout the season. They are the only team in the Top Six who play a counter-attacking style. Subsequently, they give up the majority of possession, absorb a lot of pressure and often lose the shot count in the hope that, whilst they might give away chances, by hitting their opponents on the counter, they’d create the better ones.

In order for this tactic to pay off, you need a goalkeeper playing out of their skin. Brice Samba has conceded only 31 goals from an xGC of 35.77, which is similar to Rodak’ in terms of over performance. He also tops the Championship (all those with greater than 1000 minutes played) in terms of Save Percentage with 77.8%. If he wasn’t performing out of his skin then Forest’s goal difference would far more closely represent their expected goal difference.

Adding to that, they have Lewis Grabban up front who, up until their recent run of games, was out shooting his xG by a significant amount. However, his late equaliser against Middlesbrough earlier this week was his first goal after a barren run of six games where he racked up 2.45xG, during which Forest only won twogames. Similarly to Fulham with Mitrovic, Forest are overly reliant on Grabban who has scored 35.42% of their goals.

So how’s it all going to shake out?

On the table above, I have included Brentford to show that their underlying performance numbers are actually more impressive than those of Fulham and Forest.

As you can see, despite Leeds having the best numbers for xG and xGA, they are the only side competing at the top of the league to underachieve in both metrics.

Sticking my neck out — I would say that the Top Two won’t be caught by any of the three below.

Brentford’s underlying numbers are the best of the rest, but their points gap is too big in my opinion, whilst Fulham and Forest have never looked like being consistent enough throughout the season. If they had been able to maintain the upturn in performance that brought the Top Two back into range for them in the first place, then Leeds and West Brom would have been looking over their shoulders even after a win.

Instead as the Top Two have recovered, Forest and Fulham have deservedly dropped points and it seems as if they will continue to do so.

In the case of Fulham, they are fortunate not to have fallen further back already according to their xGD. It is hard to imagine them putting together the type of run that they’re likely to need in order to break into the Top Two, particularly considering the fact that they have one of the hardest runs of fixtures remaining according to league position.

Meanwhile Forest have struggled all season in games where they haven’t been able to play their counter attacking style, summed up by the fact that they sit 12th in the league table based on home fixtures alone and they have won only twice all season when they have had more than 50% possession. Considering they need to make up a gap of eight points and an improbable goal difference swing to catch Leeds, they may well need to play on the front foot more often and it’s rather late in the season to be going through a change in play-style.

As for the title race — I think it will be a very close run thing. If Leeds can continue how they have been recently in terms of xGA performance, then they can continue to underachieve in xG performance and still have a good chance of overtaking their rivals. Should Leeds do that and also become more clinical in front of goal in the final ten games then it will be a major struggle for West Brom to keep Leeds behind them.

I’ve been saying for a lot of the season that West Brom would struggle to maintain their own over performance, but by now, it’s been going on long enough that I presume they will keep it going. If Leeds don’t have another dip in form before the end of season, then they will need to continue to be razor-sharp in front of goal or they will find themselves behind Marcelo Bielsa’s men.

Of course, I qualify all of this by saying that anything can happen in football and that nothing should be taken for granted, but it’s clear from performance across the whole season that Leeds and West Brom are much better than Forest and Fulham. It seems much more likely to me that the gap between the top two and third and fourth will be the same as it is now or even greater by the time the season ends, rather than them making it very tight. After last season’s nightmare ending, Leeds fans will be praying that I’m right.

A Leeds United blog which focuses on the tactical and statistical aspects of the game