Read on Twitter
THREAD: Using expected points (xPts) to assign value to goals:
xPts is a fun metric that's derived from xG, which uses result probabilities to gauge how many points a team "should've" gotten from the chances they created.
Result probabilities are calculated by simulating
xPts is a fun metric that's derived from xG, which uses result probabilities to gauge how many points a team "should've" gotten from the chances they created.
Result probabilities are calculated by simulating
the shots multiple times (essentially, "playing" the game 10,000 to 100,000 times) to see how many occurrences of each result there are.
Then, xPts are calculated by using the usual formula for expected outcome, which is (value of outcome)*(probability of outcome)
Then, xPts are calculated by using the usual formula for expected outcome, which is (value of outcome)*(probability of outcome)
For more on how to calculate result probabilities, check out the link here, from @FC_rstats:
http://fcrstats.com/random.html
So, using xPts, we can see how the values change before and after the scoring of any given goal in a game, and the larger the difference between before and
http://fcrstats.com/random.html
So, using xPts, we can see how the values change before and after the scoring of any given goal in a game, and the larger the difference between before and
after, the more important we can consider the goal to be.
Intuitively, think of it like this: a goal scored to make the score 1-1 is more valuable than one that turns a 5-0 scoreline into 6-0, right?
If it's still not clear, there's a nice explanation here from the guys over
Intuitively, think of it like this: a goal scored to make the score 1-1 is more valuable than one that turns a 5-0 scoreline into 6-0, right?
If it's still not clear, there's a nice explanation here from the guys over
at American Soccer Analysis (who got to this waaay before I did): https://www.americansocceranalysis.com/home/2016/6/19/goodbye-expected-goals-hello-expected-points
So, how do we use this?
Last season, Jamie Vardy won the PL Golden Boot with 23 goals and Pierre-Emerick Aubameyang came 2nd, with 22.
Applying the xPts method, we can see that even though Vardy scored 1 goal more than Aubameyang, the value of those goals was just over half of
Last season, Jamie Vardy won the PL Golden Boot with 23 goals and Pierre-Emerick Aubameyang came 2nd, with 22.
Applying the xPts method, we can see that even though Vardy scored 1 goal more than Aubameyang, the value of those goals was just over half of
what the Arsenal man's were worth. To put it crudely, Vardy 'stat-padded' a fair bit, with 12 of his 23 goals adding fewer than half a point to Leicester's tally. 3 of his goals (the 5th, 7th and 9th goals against Southampton) were worth virtually nothing, points-wise.
Compare that with Aubameyang, whose goals more often than not resulted in the addition of a point to Arsenal's total. Only 10 of his 22 were below the 1 point mark, and even then, 5 of those were above the 0.5 point mark.
This isn't a predictive stat by any means, and won't
This isn't a predictive stat by any means, and won't
give us information about the final result, so it has it's limitations. But what it does give us is a snapshot, if you will, of the goal's importance *at the moment* it is scored, and how it shifts the balance of the game.
(Data from Understat, graciously scraped by @nandy_sd).
(Data from Understat, graciously scraped by @nandy_sd).
As always, thoughts and feedback welcome.
Likes and RTs also much appreciated :)
@AshwinRaman_ @NinadB_06 @VenkyReddevil @rithwikrajendra @MishraAbhiA @abhisheksh_98 @GoalAnalysis @lgopfelix @nandy_sd @FMAnalysis @Blades_analytic @utdarena @topimpacat @amonizfootball
Likes and RTs also much appreciated :)
@AshwinRaman_ @NinadB_06 @VenkyReddevil @rithwikrajendra @MishraAbhiA @abhisheksh_98 @GoalAnalysis @lgopfelix @nandy_sd @FMAnalysis @Blades_analytic @utdarena @topimpacat @amonizfootball
