clock menu more-arrow no yes mobile

Filed under:

USL Elo Ratings - A Minor Betterment

Because you can always be more correct.

Indomitable City Soccer/Alex Leguizamo

As we continue to use Elo ratings to rank the USL, we are able to learn more about how this formula, which can be applied to all leagues, fits with the USL.  We are going to move forward, starting in week six, with one revision to a variable used in our calculations.

Elo rankings are calculated after each game based on the ratings of the teams who played, the result of the game, and a "K" value which can be used to change the weight of games (make them count for more or less).  The greater the difference in ratings, and more a team wins by, the more ratings will change.

A home field advantage is assumed to exist when calculating Elo rankings.  In the standard Elo formula, we use an addition of 100 points to the rating of the home team to reflect a home field advantage.  As all teams are starting from an 1100 rating, and the value of "K" is relatively small compared to major leagues, this means that rating will not quickly grow.  At the time of writing this, the highest rating any team has is a tie between Los Angeles and Vancouver at 1120.

Right now this 100 points creates a difficult situation, as the home team is automatically favored, regardless of who is playing.  For instance, the best team in the league visiting the worst team in the league and getting a draw will not result in points lost for the top team, and could possibly even increase its rating.  The best team should not ever be viewed more strongly after drawing a clearly inferior side.

This addition of 100 points for the home team shines light on a way to improve how ratings are calculated.  By using 50 points instead, it will prevent the away team from automatically being at an advantage.  When, through April 28th, the away team has won 40% of all matches; it shows that home field advantage does not exist to such a level where a high amount of adjustment is needed.

These corrections are very minor, and teams tended to shift in the ratings by only a few points when using a value of 50 for home field adjustment instead of 100.  The greatest change by any team was 4 points.

Thus we will be moving forward using a home field rating adjustment of 50 points as opposed to 100 points, beginning in week six.  We believe that this revision will make the Elo power rankings more accurate and reflective of the USL as a whole.