Read on Twitter
Thread:
A Finishing Model — A Project: Why Over/Underperformance in xGoals is Insufficient in Analysis
A Finishing Model — A Project: Why Over/Underperformance in xGoals is Insufficient in Analysis
One of the issues of looking at player over/underperformance in xGoals is that it does not tell you how that value is generated. In fact, it groups all types of scoring into one homogeneous group so that one-on-one finishing, heading and 30-yard shots are all seen as one.
Players are proficient in different areas of the game. For example: Player A's xG for the season is 25 and he has scored 30 goals. His overperformance is +5 | +20%.
However, if we delve into that overperformance we find that a large amount is through finishes from cutbacks.
However, if we delve into that overperformance we find that a large amount is through finishes from cutbacks.
Also, a large amount of those cutbacks have come from the right-hand flank, creating by one or two select players.
If we remove these from the sample we find he is underperforming in one-on-ones, individual chance creation, long shots, cutbacks from the left and general crosses.
If we remove these from the sample we find he is underperforming in one-on-ones, individual chance creation, long shots, cutbacks from the left and general crosses.
However, the data will show an overperformance of +5 | 20% and credit the player for ‘good’ finishing despite the fact that their overall skillset in finishing is deficient.
This impacts scouting massively as a team may not create the specific chances Player A converts. They might not have prowess down the right-hand side but, instead, have it down the left and do not have the midfielders to create that space for cutbacks but, instead, crosses.
Also, Player A may be playing in a game and the midfielder who creates space for that cutback due to his proficient playmaking may be injured despite the winger who assists him being available. As a result, that opportunity doesn't come and the player wastes chances as a result.
After the game his total overperformance drops by –1.5 | –6.8% to +3.5 | +13.2% and it looks like “they were unlucky” when, in reality, they were most likely going to struggle as they are not proficient from other chance creation.
This is one mere example of something that is clear when looking at players. Their over/underperformance is driven by finishing chances but chances are created differently and players excel at different things.
This is something analysts must highlight.
This is something analysts must highlight.
This is why I have spent time developing a finishing model and it is where this tweet on Greenwood below came from:
My model looks at three things:
• the chances a player receives
• the conversion of said chances
• how the player converts them
The third is a biological analysis on which parts of the foot the player uses coupled with muscle activation. This massively impacts conversion.
• the chances a player receives
• the conversion of said chances
• how the player converts them
The third is a biological analysis on which parts of the foot the player uses coupled with muscle activation. This massively impacts conversion.
One of the important elements under the first aspect is whether a player is contested or not when they shoot which has a massive influence on whether a player scores. Likewise, it factors in spacing when the player shoots. The next tweet has an example:
From my sample of top scorers, Kylian Mbappé scores a higher percentage of his total goals than most players in this area of the box — a high probability zone. Also, a large amount of them are uncontested and a good amount of them are one-touch finishes.
Now, someone might say “he plays on a great team that creates big chances for him”. However, a significant percentage of these goals have actually come as a result of Mbappé's movement quite like Cristiano Ronaldo's in-box finishes.
This is something my model looks at.
This is something my model looks at.
I am still working on it and the first piece of analysis will be on a selection of top scorers from the top five European leagues. It will highlight the types of chances they receive, their conversion, issues of spacing and defender, biological analysis and more.
This was just a short thread to introduce my plans with a short explanation as to why it is important to look at when delving into goalscoring and players either over or underperforming their xGoals.
End of Thread.
