The American data based news website FiveThirtyEight showcases their football predictions for most of the leagues in Europe and across the world here. Though these predictions are at times divisive, the models they use to estimate individual match performances are founded on strong principles. They use three metrics to evaluate a team’s performance after each match shot-based expected goals (xG), non-shot expected goals (nsXG) and adjusted goals. The first metric among these three are quite popular in footballing community with xG going mainstream even on BBC’s Match of The Day.
FiveThirtyEight explains these three metrics like this:
xG are an estimate of how many goals a team “should” have scored, given the shots they took in that match. Each shot is assigned a probability of scoring based on its distance and angle from the goal, as well as the part of the body the shot was taken with, with an adjustment for which specific player took the shot.
nsXG are an estimate of how many goals a team “should” have scored based on non-shooting actions they took around the opposing team’s goal: within an area slightly larger than the 18-yard box.
adjusted goals counts for the conditions under which each goal was scored. For adjusted goals, we reduce the value of goals scored when a team has more players on the field, as well as goals scored late in a match when a team is already leading.
Summarizing, we have collected three metrics which evaluate not only a team’s performance in a match but also the actual goals scored adjusted to in-match context. Using these three metrics, I re-calculated individual match results and built a Performance Adjusted League Table. The procedure used to built the table is described below.
Since we have three metrics reflecting a team’s performance in a match, I decided to use a composite metric of average goals per match which is an average of xG, nsXG and adjusted goals. Once the average goals were found, I assigned a win, loss or a draw based on the difference in average goals in a match. I used a value of 0.2 as the threshold for decision.
To elaborate, for a match of Team 1 vs Team 2, if average goals of Team 1 was greater than Team 2 by 0.2, Team 1 wins the match. If average goals of Team 2 was greater than Team 1 by 0.2, Team 2 wins the match. If neither of the above are true, then the match is a draw.
Using this criteria, all the matches in 2020–2021 season of the Premier League was re-evaluated and Performance Adjusted League Table was calculated. The table updates automatically after each matchday. Please check it out regularly to see how your team is doing. The interactive table can be viewed here.