The Case of the Fulham FC and the Missing Goals

Fulham have struggled for goals this season, with just 15 scored in the league so far (writing after the Man Utd game). This puts them tied with West Brom for 17th in the goal scoring charts and indeed only Burnley and Sheffield United have scored fewer.

But the stats suggest Fulham should be doing much better than this, after all, they rank a more respectable 13th for total shots and 14th for Expected Goals.

For those unfamiliar with the concept, Expected Goals (or xG) is a measure of how many goals we would expect a team to score based on the quantity and quality of the chances they generate.

There are different companies which model expected goals, but looking at the model used on the FBRef website, which is supplied by StatsBomb, we see that Fulham have been expected to score 20.1* goals so far, but, as stated above, they have actually only scored 15.

This means that, so far this season, Fulham have only scored 75% of the goals that they have been ‘expected’ to. This ‘xG Conversion Rate’ sounds bad, but how bad is it?

Well, only two teams have fared worse in converting expected goals into actual goals this season; Burnley and Sheffield United (converting each at around 55%). The xG Conversion Rate for all Premier League teams is shown in the chart below.

Looking longingly at the other end of the xG Conversion rankings, we see Southampton, with expected goals of 17.3 (i.e., about 3 less than Fulham) but actual goals of 26 (a whopping 11 more than Fulham), an xG Conversion Rate of 150%!

So, the obvious question then is why are Fulham doing badly at turning good scoring chances into actual, real life goals? The numbers tell us that Fulham are missing around 5 goals (5.1 to be precise), so let’s see if they can tell us anything about why they are missing?

The focus of this article is then to try and say something about how and why Fulham are missing these goals.

Our investigation starts, of course, at the penalty spot. Fulham had a horrid run this season missing 3 pens in a row at critical moments. At present, Fulham have converted 2 of 5 spot kicks in total this season. Expected goal models tend to estimate the probability of scoring from the penalty spot at 76%, so Fulham would expect to have scored 3.8 goals from their 5 penalties. With only 2 actual goals scored we can immediately identify the source of 1.8 of our 5.1 missing goals. But that still leaves 3.3 goals unaccounted for…

Are Fulham just bad at shooting?

One possibility is that Fulham are simply not very good at shooting as a team. This has certainly been the case when it comes to penalties, but is it the case from open play as well?

Fortunately, we can apply some data to this question. The expected goal numbers I have mentioned above are based on a ‘pre-shot’ model; this means the expectations are based on the position (as well as other factors) from which each shot was taken.

But we can also look at a ‘post-shot’ expected goals model which tells us how likely a shot is to result in a goal based on its placement in the goal and its power.

Fulham have accumulated total ‘post-shot’ expected goals of 18.4 this season. This is 1.7 goals short of the number of ‘pre-shot’ expected goals (20.1), but again if we think about penalty taking, recall that Fulham have had two penalties (with combined pre-shot xG of 1.5) which missed the target entirely (thus generating ‘post shot’ xG of nil).

This means that penalties explain nearly all of the difference between Fulham’s post-shot and pre-shot expected goal totals, in other words, excluding penalty taking, Fulham’s shots have been pretty much exactly the quality that we should expect, given the chances from which they were generated.

So that is interesting, poor shooting from open play does not seem to explain Fulham’s missing goals. Fulham’s open play shooting is, in fact, extremely average (but no worse than that – phew)!

Is it good opposition goalkeeping?

If we have deduced that Fulham are taking shots of roughly the quality that models would expect, then we can assume that the reason for the missing goals is that an unexpectedly high number of Fulham’s shots are being saved. Does then the data support this assumption?

It does seem so, yes! The chart below shows, for each match this season, a green bar where opposition goalkeepers have done better than ‘expected’ by the Post Shot xG model, and the red bars show where they have done worse than expected.

The orange line tracks the cumulative performance of opposition goalkeepers with regard to whether they are stopping more or less than models predict.

We can see that up to, and including, the game against Everton, there was a fairly random mix of outcomes, sometimes the bar is red and sometimes green, and the cumulative measure of the orange line shows that Fulham are scoring roughly the same number of goals as we would expect, given the shots being taken.

From the Leicester game onwards, however, the picture seems to change in an alarming way, there is a run of 9 games where opposing keepers are typically outperforming the model, saving more shots than we would expect.

Only in the Newcastle game is this reversed (then only courtesy of an own goal) with the aggregate of goalkeeper overperformance rising to 3.6 – a good match to the number of missing goals we were trying to locate in our investigation.

To put this in simple terms; since the Everton game, for some reason, goalkeepers have been doing unexpectedly well in saving Fulham’s shots.

Why is this happening?

If goalkeepers have suddenly started saving all of Fulham’s shots, it is reasonable to wonder why this might this be?

It could be simple bad luck (from Fulham’s perspective) i.e. we just happen to have had a run of 9 or so games where goalies have had great days against us, this is perfectly possible over such a small sample size.

This would give cause for optimism that a change of luck could be around the corner and the goals might start flowing again.

But one can’t help but think about the current striking dilemma in the team, namely finding a role for Aleksandar Mitrovic.

Mitro has not been starting recently, indeed his loss of a starting spot in the team seems to correlate quite closely with the rise in missing goals in the chart above.

Looking in more detail, we see that of the 3.3 missing non penalty goals we are trying to hunt down, 3.1 occurred in games where Mitro did not start!

Does this mean that answer to our mystery a simple matter of missing our elite finisher Aleksander Mitrovic?

Unfortunately, it is not so simple. If we look at the performance (actual goals) versus (non-penalty) expected goals of Fulham’s 4 main offensive contributors (Mitrovic, Lookman, Cavaleiro and Decordova-Reid) we see that Mitro is the player most guilty of failing to convert chances, scoring 1.8 goals fewer than the models would predict from open play (see chart below).

What seems to have happened this season is that, in the first 10 or so games, Mitrovic was underperforming his xG, but Decordova-Reid was overperforming his, as a result Fulham’s overall performance was about neutral.

Scott Parker’s tactical changes then saw Mitro’s game time limited, but also saw Decordova-Reid deployed in a more defensive capacity (often at wing-back) which limited his goal involvement.

Instead, it was Cavaleiro who became the main focus of Fulham’s attack and, as you can see in the chart, he has been quite unsuccessful in converting chances into goals.

Conclusions; what have we learned?

To summarise the findings in our investigation:

  • Fulham are converting only around 75% of their ‘expected goals’ into actual goals and are therefore missing about 5 goals this season.
  • Of these 5 goals, about 1.8 can be attributed to Fulham’s underperformance from the penalty spot,
  • But outside of penalties, Fulham seem to be taking shots that are as good as one would expect, given the chances they are creating
  • Instead, the remaining missing goals seem to be accounted for by run of good goalkeeping performances against Fulham (starting with the game against Leicester)

From the final point I think there are two possible conclusions either:

1) there is something about Fulham’s shooting during recent games (perhaps associated with Cavaleiro as the main attacker or with Fulham’s counter-attacking style) which is causing post-shot xG models to consistently over estimate the quality of Fulham’s shots, or,

2) Fulham have just had a bit of an unlucky run…

My personal view is that the latter option is the most likely, which is encouraging as it implies that a swing in fortune towards Fulham (or just back to the mean) would yield a substantial increase in goal productivity.

I am therefore optimistic that Fulham’s goal scoring will improve in the coming gameweeks, and of course the xG conversion rate is something I will continue to monitor going forward, and provide updates on as necessary!

*The actual figure quoted on FBRef is 19.3, but they give Ivan Cavaleiro no xG for his missed penalty against Everton. This is because he double-kicked the ball, and therefore did not actually manage to take a legal shot. However, I find this unhelpful for analysing Fulham’s season, so have re-inserted the xG for this penalty into Fulham’s total, bringing it up to 20.1.

One thought on “The Case of the Fulham FC and the Missing Goals

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: