Opponent-adjusted margin in the Premier League, and predictions for the end of the season

Introduction

Opponent-adjusted metrics have been discussed at length on this blog before, beginning with a posts on college football in 2012, looking at opponent-adjusted record, margin and then using the opponent adjusted margin to evaluate teams which teams had over- or under-acheived. After this, the 2012-13 Premier League season was considered, first looking at managerial changes and, following that, looking at if "form" exists.

This post is to be the first of a series throughout the season, tracking opponent-adjusted margin in the Premier League and what it means for a team's prospects throughout the season.

Method

The method for calculating opponent-adjusted margin is documented in this previous post. Following the calculation of the average opponent-adjusted margin, the standard deviation of performance for each match is calculated. Using this information, a Monte Carlo method is used to simulate the games remaining to be played in the season, providing a final league table, based upon each team's results to date.

Results

Shown below is a table with, for each team, the average opponent-adjusted margin for each game played so far this season, the standard deviation of individual match performances and a number of probabilities: the probability of winning the league, that of finishing in the other Champions League qualifying positions, that of qualifying for the Europa League and finally that of being relegated:

Conclusions

The first thing to be said is that these probabilities should not be taken as actual probabilities of a team's performance in this season's Premier League. The sample is so small (3 games) that far too little is known of each team's actual ability. Instead, these are probabilities of these outcomes if each team continues to achieve results at the same level at that seen so far this season.

Even so, it is still interesting to note some outcomes so far:

• Liverpool's 100% record puts them at the top of the table, as expected
• Despite taking only 4 points from 3 games, the strength of the opponents (a loss against #1 Liverpool and a draw against #2 Chelsea) combined with the margin of victory against #13 Swansea City puts Manchester United in 3rd
• #4 Aston Villa are higher than might be expected, given that they have only accumulated 3 points from 9 games. The two losses, however, we against #7 Arsenal and #2 Chelsea
• Although #6 Manchester City are 3rd in the league, a loss against #10 Cardiff City and a best win against #11 Hull City means their opponent-adjusted performance is 6th best
• #18 Southampton's 10th place position in the Premier League table flatters them, as they have enjoyed a relatively easy start to the season
• The opposite is true of #11 Hull City, who lie in 18th in the Premier League table
• #19 West Brom and #20 Sunderland are 20th and 19th, respectively, in the Premier League which should cause concern

As mentioned above, these metrics will be tracked throughout the season and the results published. More data is needed to put any great confidence in the probabilities, but they should be useful later on.