Thread by @Reuser5

A (LONG) STATS THREAD - Expected Goals on Target (xGOT)

Those of a nervous disposition when it comes to expected stats should probably look away now. 😬

I'm going to discuss a concept that is new to me - xGOT. And how it might help FPL managers.

(1/15)

Jokes about Game of Thrones and "Swords and Shields" aside, xGOT stands for "expected goals on target".

Whilst xG looks at the quality of a chance, xGOT looks at the quality of the actual shot. Is the shot high quality (in the corner of the goal) or low (centre of goal)?

(2/15)

Needless to say, it only looks at shots on target, as a shot off target has a near 0% chance of leading to a goal.

In a sense therefore, it is similar to the FPL metric "Shots on Target (SOT) - but looks at the sum quality of the shots rather than the number of shots.

(3/15)

xG merely shows us which players get in the best scoring positions the most often.

Auba is a great example - his very high xG shows that he is fantastic at getting good chances.

But xG alone doesn't tell us much about what he does with those chances.

(4/15)

If we compare his non-penalty xG with his non-penalty xGOT however, it is clear that his finishing doesn't match his ability to make good chances:

NPxG = 19.16
NPxGOT = 14.28

I have since learned that this difference (-4.88) is called the SGA (Shooting Goals Added).

(5/15)

SGA is a measure of how much value a player has added to (or removed from) their chance leading to a goal. In other words,it can be used as a proxy for finishing skill.

The higher (or more positive) the SGA - the better finisher.
The lower (or more negative) - the worse.

(6/15)

I took every player from 18/19 and filtered (largely arbitrarily) for players who:

- scored 7 or more goals
- played more than 1000+ minutes
- played in the 19/20 season

I looked at both their NPxG and NPxGOT and therefore was able to determine their SGA.

(7/15)

17 players had an NPxGOT that exceeded their NPxG by at least 10%:

JWP +60%
Fraser +52%
Martial +50%
Pedro +44%
Son +42%
Felipe A +37%
Sane +33%
Eriksen +26%
Ayoze Perez +20%
Salah +19%
Lukaku +12%
Moura +11%
Mahrez +11%
Deulo +11%

(8/15)

These 17 players can be argued to be last season's "elite finishers" - players who added value to their xG through the quality of the placement of their shots. These aren't players who "got lucky" - they are chances that are well-taken and more likely to lead to goals.

(9/15)

And the "worst finishers"?

Well -there are plenty across the game - but if we want to look at the most negative SGA within the group of players who scored more than 7 goals last season - we have some very interesting names.......

(10/15)

12 players had an NPxGOT that was lower than their NPxG by at least 10%:

Ings -39%
Jesus -29%
Aubameyang -25%
Jimenez -23%
Zaha -19%
Wood -19%
Vardy -16%
King -15%
Hernandez -15%
A Barnes -13%
A Gray -13%
Aguero -11%

(11/15)

It is important to note that this doesn't makes these 12 players bad FPL picks - in many cases their sheer volume of shots and/or quality of chances can compensate for the relatively poor placement of their shots.

(12/15)

But it does indicate that in some of cases, we need to bear in mind that their xG does need to be reconsidered in the light of their relatively poor placement of their shots.

(13/15)

If you have made it this far - congratulations. Hopefully I have been able to convey a stat and concept that is new to me, and hopefully useful to us all.

xGOT can be used to determine the quality of a shot and therefore the finishing ability of a player.

(14/15)

All statistical research for this thread was conducted using @FFScout's excellent membership area and custom tables.

A good article on the topic can be found at: optasportspro.com/news-analysis/…

Thanks to @rogue_wee and @StatOnScout for acting as a sounding board for this thread.

(End)

Expected Goals on Target (xGOT) - AN ADDENDUM.

Thanks to everyone for the conversations generated by this thread. So many great points raised in parallel discussions.

With this in mind, I'll be summarising key points in an addendum.

First up - Opta definition of xGOT.

(1/n)

Plenty of discussion regarding why long range shooters come out more favourably in the analysis.

Thought: if we replace the term "finishing skills" with "shooting accuracy", is that a better reflection of what we are looking at?

(2/n)

Whilst xGOT is a new metric - I have since discovered that it appears to be have been applied retrospectively all the way back to the 15/16 season, so our potential sample is much larger than previously thought.

I intend to re-run the analysis for all available seasons.

(3/n)

Some great discussion whether there is a better way of calculating finishing ability from xGOT and SGA.

(A revised equation is suggested a few posts down in this thread)

https://twitter.com/hsorsky/status/1150380297927757825

(4/n)