Leeds United, xG & Conversion — a problem Marcelo Bielsa can’t seem to fix

It was the same old story on Saturday: Leeds had the overwhelming share of possession, outshot their opponents by 17 shots to 6 and won the ‘xG battle’ comprehensively. Inexplicably, they lost.

It doesn’t seem to matter anymore what the scoreline was or who the opponents were; this is simply the story of almost every game Leeds United have dropped points in over Marcelo Bielsa’s tenure as Leeds United manager.

In fact, the only loss Leeds have suffered this season where the xG scoreline had them as ‘fair losers,’ was the 2–1 loss to Millwall at the New Den, which was largely due to a wrongly awarded penalty. Subtracting the 0.76xG that accounts for the penalty, Leeds would have had the greater xG in that game as well.

We’re very aware at All Stats Aren’t We that not everybody is a fan of xG. After Spurs’ 2–0 win against Man City on Sunday, journalist Samuel Luckhurst called it ‘the most joyless thing ever to happen to football’ and we’ve had our fair share of replies on Twitter criticising the use of the metric.

However, the fact is that it remains one of the best tools we currently have to measure chances quality and thus give a more accurate story of what happened in a game rather than reverse-engineering it from the scoreline.

Simply taking shot count alone isn’t quite enough. For instance,Team A could take 15 shots from poor positions due to not being able to play through defensive lines. Meanwhile, Team B might take only two shots on goal, both from central areas between the 6 yard box and the penalty spot.

Looking simply at match stats it might seem like Team A deserved the win, when the truth is that Team B created the better opportunities to score. xG simply gives us a value to express that.

Sample Size, Stupid!

More than simply analysing individual games — which can be victims of wild variance — xG is more helpfully used as a means of seeing long-term trends and making predictions about where a team might finish in the league.

For example, xG tables have had Leeds clear at the top of the Championship throughout the season and many models had forecast Leeds to win the league comfortably due to their ability to consistently create more than their rivals.

Whilst Leeds wouldn’t have to track their xG perfectly in order for this to be the case, the theory is that even a side underperforming their xG would still score enough goals to comfortably win the league as long as they continued to generate xG in the way they have throughout the season.

There is also a theory in football analytics that finishing is effectively a myth — i.e. consistently getting in good positions is far more important than ability to finish due to high-level professional footballers being roughly at the same level of ability and thus having similar conversion rates. If you’d like to read more about that theory, we’d thoroughly recommend this article from Bobby Gardiner.

The Exception that Proves the Rule

All that being said: Leeds United’s finishing is so bad that it seems to compromise that theory and even the idea of using xG to forecast table positions.

Leeds’ under-performance on their xG cost them automatic promotion last season and they are in serious danger of it doing so again as their conversion rate is currently lower than last season and they are under-performing their xG to a level far greater than any other team in the league in 19/20.

To illustrate this point, here is a table of some of Leeds’ shooting stats over the last two seasons:

Here are the Top Six from 18/19 and their performance against xG:

Here is this season’s Top Six and their performance against xG:

As you can see, Leeds stand out as atrocious in comparison to their closest rivals.

There is still time for them to improve on this season’s xG performance but it’s a lot to ask for them to improve a lot when there has been no suggestion lately of their finishing improving.

In fact, they are incredibly fortunate that their rivals this season haven’t been as clinical as last season’s Top Six or they could find themselves in a more difficult situation than they’re already in.

A question we often get at All Stats Aren’t We is: ‘Why are Leeds so bad at finishing?’ It’s very hard for us to answer this question but I’ll attempt to look at some of the things people have suggested as reasons.

Marcelo Bielsa’s System

One thing that has cropped up a few times is the idea that Bielsa’s system is the problem.

The theory goes that, although the system creates more than any other team in the league, perhaps it’s so physically demanding on players that by the time it comes to finishing the chances created, the players are exhausted and their composure is therefore poor.

My first instinct is to dismiss this out of hand — there’s plenty of examples of teams that play high-intensity football who manage to finish clinically, or at least track their xG.

Case in point: Norwich were third in the league for pressing in 18/19 (behind only Leeds [1] and Derby [2]) and 5th in the league for passes per minute.

However, they scored 93 goals from only 79.04 xG and 659 shots giving them a conversion rate of 14.11%. These are obviously far more impressive figures than Leeds’ and were ultimately the reason why Norwich won the league and Leeds finished in third.

In order to look into this a little further, though, I looked at figures from Marseille’s 2014/15 and Athletic Club’s 11/12 and12/13 seasons as they were the last three full seasons Bielsa managed at club football before he came to Leeds.

I wasn’t able to get xG figures for those seasons but their conversion rates were as follows:

The average conversion rate of all teams in those seasons varied between 10.3% and10.9% and, in the case of two out of the three seasons listed here, Bielsa’s teams were above average (Athletic Club 11/12 & Marseille 14/15).

During his second season at Athletic, they were below average in terms of conversion rate. That said, there was a major issue as the club’s star striker Fernando Llorente was frozen out of the side due to contract issues and it subsequently turned out to be Bielsa’s last season at the club, so it seems better to focus on Athletic’s 11/12 season & Marseille’s 14/15 as neither of those were affected by off-field issues.

As such, it’s a fair conclusion that Bielsa’s system isn’t a factor in the low conversion rates.

Is it All Patrick Bamford’s Fault?

A common theme on Twitter after every Leeds loss is a barrage of tweets apportioning blame towards Patrick Bamford.

This piece is by no means a defence of his finishing, as it has been very poor as you will see in the table below. However, conversion rates across the whole team are lacking.

Of course, the common argument is the striker has to score goals, but Bamford is comfortably the top scorer in the Leeds team.

One would expect other players to be troubling that status considering how many chances he’s missing but they simply aren’t weighing in with enough goals. Of the Top Six, only Leeds and Nottingham Forest have fewer than 3 players on 5 goals scored at this stage of the season:

As we can see here, Bamford’s xG performance is nothing short of woeful and is the worst in the league. But he is the only Leeds player with a conversion rate above the league average of 11.36% (minimum 20 shots taken).

Unfortunately, he is way down on the average conversion rate for strikers who have played over 1000 minutes, which is 15.87%. At that conversion rate, he would have 15 goals by this stage. Notably, though, that would still be underperforming his xG, which speaks volumes about the quality of chances he’s had.

Leeds Shoot from Bad Positions

The third theory that gets floated our way is the idea that Leeds have high xG because of the cumulative effect of lots of low-value shots. Again, this is easily refuted.

Leeds rank sixth in the league for xG per shot with a value of 0.121. Meanwhile, West Bromwich Albion are the top scorers in the league and sit nine places below Leeds in these rankings with an xG of 0.111 per shot.

Additionally, all chances over 0.3xG are considered ‘Big Chances’ and are defined by Opta as chances that are ‘reasonably expected to be scored.’

Leeds create significantly more of these chances per 90 minutes than anybody else in the league. As always, their conversion rate is the problem, as this graphic demonstrates:

Leeds have created 78 of these ‘big chances’, whilst West Brom sit behind them on 66. However, West Brom’s conversion rate is so much better than Leeds, who have scored from 25 of them. West Brom have scored 28. If Leeds had the same conversion rate as West Brom they would have scored 8 more goals.

If a visual representation of these chances is more helpful, here are the shot maps for Bamford and Harrison, Leeds’ top shot takers this season:

Bamford’s Shot Map
Harrison’s Shot Map

As above — there are an alarming amount of efforts taken from inside the six-yard box that haven’t been scored or even hit the target.

The Players aren’t as Good as their Top Six Rivals

This is the take that I have been most sympathetic to. One can’t help but feel that Bielsa would blow away the league with forwards like Brentford’s Said Benrahma, Bryan Mbeumo and Ollie Watkins, West Brom’s Matheus Pereira, Grady Diangana and Charlie Austin and Fulham’s Ivan Cavaleiro, Anthony Knockaert and Aleksandar Mitrovic.

It’s easy to say that Leeds have occupied a lofty position in the table through Bielsa’s tenure purely because of the system, rather than the players themselves. However, with the amount and quality of chances created, Leeds don’t have to have the best finishers in the league, they merely have to be average.

As we’ve already ascertained, they haven’t been. This doesn’t make any sense when considering the quality of players — Pablo Hernandez, Helder Costa, Mateusz Klich, Jack Harrison and, yes, even Patrick Bamford, are all players considered well above average: even elite players in the division. Yet somehow, when it comes to putting the ball into the net, they’re not.

Importantly though, this doesn’t seem to be the case for these players when not playing for Leeds United. Costa matched or out-shot his xG in every season in a Wolves shirt, whilst Bamford overperformed his xG by three goals in his last season at Middlesbrough.

It’s All in the Head

So that leads us to the intangibles.

Perhaps it’s become a self-fulfilling prophecy that Leeds miss chances and drop points due to poor finishing. During Leeds’ seven-game-winning-run, the team seemed confident in front of goal, scoring 15 goals. Since then, they’ve managed only 14 goals from 9 games with 11 of those coming in three games.

It seems that, as Leeds miss chance after chance, the likelihood of them scoring drops; rather than feeling like a goal will inevitably come, confidence drains away.

There is a feeling that, if Leeds score early, they’ll grow in confidence and score more. But with every miss, the players and certainly the fan-base, think ‘It’s happening again!’

Can it be fixed?

The honest answer is that I simply don’t know.

We’ve been saying all along that Leeds finishing must surely come up to the average conversion rate in time and the most important thing was that they were creating chances regularly. However, time is running out for it to happen this season and it’s been this way for pretty much a full calendar year.

It feels like things now rely on Bamford hitting a hot-streak similar to the one he managed for Middlesbrough in 17/18, where he scored nine goals in seven games or for new signing Jean-Kevin Augustin coming quickly into the team and hitting the ground running.

There are two questions to ask here: firstly, will Augustin be brought into the side in enough time to make an impact; and, secondly, can he finish better than Bamford?

On the former: it’s very hard to second guess Marcelo Bielsa but, if the lack of Augustin on the bench for Saturday’s loss against Wigan is anything to go by, we can expect him to not be rushed into the side regardless of how Bamford is finishing. One would hope that he is used as a weapon from the bench if goals are needed, though.

On the subject of his finishing: he has 23 goals from 133 shots (17.29% conv. rate) and an xG of 20.85 across all of his appearances in first-team football. These are the numbers of a very good finisher and are a vast improvement on Patrick Bamford’s current figures.

One thing that is for sure is that it’s unlikely to be an easy ride from this point onwards. It would take a huge upturn in finishing for Leeds to suddenly start blowing teams away, but they certainly create enough to do that, so it’s not impossible.

It’s now make or break. Either Leeds — as a whole team, not just one man — find their shooting boots or automatic promotion will slip through their grasp once again.

You can follow Josh Hobbs on Twitter @JoshAHobbs.

If you enjoy this content and want to help us create more and better pieces, why not consider signing up to our Patreon page to get bonus material?

If you have any interest in contributing to this blog, get in touch with us @AllStatsArentWe.

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