RATINGS FIXES   (State Scaling)


The purpose of this change was to finally get every state to fall exactly where it should...slotting in against the rest of the country.  As those who follow the site closely know, the ratings are tremendously accurate in any given area, as there is so much data to go on.  That hasn't always held true outside of state lines though.  For example, sometimes you'll see a team score an impressive upset win against an out-of-state national power, and of course the winning team (and their state as a whole) get a bump out of that, but it can sometimes appear to the naked eye as if it wasn't enough of a bump.

It turned out that there was a relatively easy fix to this.  It was as simple as adjusting the way a state as a whole is rated AFTER the ratings have been created.  So, if a state has played a large sample size of out-of-state games, and consistently done better than ratings would say was expected, then obviously that state needs more recognition.  We're not talking about a state that has only played a couple of out-of-state games here, but say for example that a state has played 30 out-of-state games and averaged winning those games by 2.5 points more than the ratings would suggest (discarding blowouts).  That state simply then gets a +2.5 point bump in terms of where they slot in against the rest of the teams in the country.

As an example, IL/GA/UT/AZ were the main states that needed bumps in 2016, as it turns out...but of course, the list of states changes from season to season.  Each of those states mentioned were anywhere between a point and three points lower than they should have been.  And needless to say, a few states were rated too high as well, and we've made those changes also.  We're not talking about changes that are particularly noticeable to the naked eye, necessarily, but the overall ratings are definitely helped by this change, as the states fall in line with each other right where they should.

Fortunately, none of this will be an issue going forward.  States are automatically scaled now-- adjusted for how much better/worse they've done against out-of-state competition than would have been expected.  So again, not a problem going forward and also, we've made the change retroactively to previous seasons as well.

One interesting thing to note is regarding Texas.  While it's still "king", the edge the state holds isn't nearly as pronounced as it used to be.  This in large part has to do with the fact that there are football-factory type programs popping up all over the country now, so a big chunk of the national talent isn't located in one state, as it used to be.  We won't again see the days where nearly half the teams in the top 25 by an objective rating system were from Texas.

A statistical confirmation of the lessening of Texas's dominance can be found in the narrowing of their lead in our state ratings.  While they continue to lead the pack, and have been the #1 state every single season we've been covering the whole country, from 2003 to present, the gap has become razor-thin the last three years.  Here's a list of the margin of Texas's lead over the #2 state, by year, in terms of the average power rating of the top 50 teams in each state:

2003 2.5
2004 8.1
2005 11.0
2006 9.2
2007 8.2
2008 3.6
2009 6.2
2010 4.5
2011 7.3
2012 2.7
2013 3.3
2014 0.4
2015 0.6
2016 0.2

Another indication of the evolving relative dominance of Texas is the number of national top 25 teams an objective mathematical system awarded them by year:

2003 4
2004 10
2005 12
2006 9
2007 9
2008 8
2009 8
2010 7
2011 5
2012 5
2013 5
2014 4
2015 5
2016 2