2014 - MARS and ChiPS Ratings After R1

/

As I've written about before, MatterOfStats has an additional team rating system this season, called ChiPS, and designed so that the difference in the ratings of any pair of teams is a direct assessment of the margin of victory you'd expect to see were they to meet on a neutral ground.

Since it's a little early in the season to be calculating Colley, Massey and ODM team ratings and comparing them with MatterOfStats' traditional team rating system, MARS, I thought I'd instead this week compare MARS Ratings with ChiPS Ratings.

The table below provides each team's ChiPS Rating at:

  • the end of the 2013 home and away season
  • the end of the 2013 Finals
  • the start of the 2014 season (after applying the between-season regression to the mean adjustment, which sees teams carry over 61% of their end-of-season Rating, to which is added 390 Rating Points)
  • the end of Round 1 of 2014

It also provides each team's MARS Rating at the end of Round 1. (For the equivalent team MARS Ratings at the end of season 2013, see this final blog from last year.)

One of the things that's apparent from this table is how much more variable on the basis of a single result are team ratings under the ChiPS methodology than we've seen previously under MARS: four of the Round 1 contests moved ChiPS team ratings by over 4 Rating Points for the victor and for the vanquished, an almost inconceivable increment in terms of MARS Ratings.

In Fremantle's case, the ChiPS Rating change was sufficiently large to precipitate a 4 spot jump on the rankings, from 6th to 2nd, and in West Coast's circumstance it was enough to drive a 3 spot climb from 15th to 12th.

Despite these movements and a number of smaller ones, the correlation between the team ratings under ChiPS and under MARS at the end of Round 1 is very high at +0.977, and the team rankings are broadly consistent. The largest differences is rankings are for:

  • Richmond, ranked 4th by ChiPS and 7th by MARS
  • Sydney, ranked 7th by ChiPS and 4th by MARS
  • Carlton, ranked 11th by ChiPS and 9th by MARS

For no other team is the ChiPS ranking more than 2 spots different from the MARS ranking.

Lastly, whereas the difference in ChiPS Rating is, by design, the best estimate of the difference in team abilities measured in terms of points, empirical analysis using the Nutonian Eureqa product suggests that about 0.722 times the difference is the best equivalent measure using MARS Ratings.