Edouard – Over Performance or Massive Potential?

Celtic’s scoring prowess reduced this season. Celtic scored 126 goals over 61 matches (2.07 per game) compared to 150 over 59 last season (2.54 per game). Edouard scored 11 from 14 starts. Impressive? Yes! But that’s a lazy stat and here’s why.

The Glossary buries lack of clarity from any distance into the bottom corner, off either foot.

Total Goals

The spread of goals by competitions is as follows.

Straight forward? No.

Goals per 90m

Removing differences in minutes played gives a truer reflection of a players scoring propensity. Goals per game has reduced meaningless in any era since substitutes were introduced. Forward players are most often substituted. You need to factor minutes on the pitch to assess and compare goal scorers. And of course it gives us a very different list because of that.

Secondly I remove penalties as this is a very specific and different skill. Irrespective of whether the penalty taker is Messi, Ronaldo, Boyata or Ciftci, the xG is about 0.769. We should measure penalties as a specific skill – like splitting heading duels from other challenges.

So, here are the Non penalty Goals scored per 90m (NPG90):

Once you remove penalties, Dembele (4 penalties scored) and Sinclair (2) drop down the list. Edouard is the standout with 0.756 goals per 90m. Often used as a substitute he only started 14 matches, scoring 11 goals. This is clearly a very promising number but as you can guess, not the whole picture.

xG

Our old friend Expected Goals (xG) really tells you what you deserved to win rather than what you actually got based on the quality of the chances. Here is the comparison between actual goals scored and xG with penalties added back. Those at the top of the chart scored more goals than xG predicts. Those at the bottom less.

As a squad Celtic scored 4.04 less goals than expected over the season. This compares to last season when I believe time will show they had “one of those seasons” when many chances went it that normally wouldn’t and the xG was massively outperformed. I can only compare SPFL games but Celtic scored 106 against an xG of 95.222. A 10% over performance last season compared to a 3% under performance this. The bad news is I expect this season to be nearer “normal” than last.

The second observation is that the players at the top of the table – those that over performed versus their xG – are the ones perceived to have had a good season. McGregor, Forrest, Ntcham and Edouard are all perceived to have performed well. Conversely, Sinclair, Dembele, Rogic, Armstrong, whilst not considered failures, are perceived to have had slightly more inconsistent seasons than the aforementioned.

What is fascinating to consider is that if the numbers were reversed would our perceptions have changed? I raise the question as it seems quite small margins (an extra 2-4 goals scored) might alter our perception of what “good” looks like. Over performing your xG is not always about luck of course – consistent excellent finishing might be at play. Also, the likes of Forrest, McGregor, Ntcham benefitted from extended runs in the team and few injury interruptions. That means rhythm and timing and confidence are all improved. For the likes of Armstrong, Rogic, Sinclair, Dembele, being in and out the team and recovering from injuries dents the same attributes. The fitter and more confident and match ready you are the luckier you are, maybe?

So, will the patterns be sustained over the long term? The point is we can expect regression to the mean over a long period.

What that means is we should be cautious regarding Edouard. He out performed his xG. It’s a small sample. Will he regress to the mean? Or, will an extended run in the team see him over perform to a higher degree?