Special #1 Explanation

posted Oct 3, 2010, 10:50 AM by Ted P   [ updated Oct 3, 2010, 11:02 AM ]
I wrote this to try to explain why the ratings came out the way they did, so I think it might help people understand how it works.  If not, just disregard.

Without trying to get too technical, there are two basic steps to my ratings, rating teams as opponents and then using those ratings to calculate the overall ratings.  When we’re only talking about wins, your rating directly corresponds to the ratings as opponents of the teams that you beat.  By telling you how teams are rated as opponents, I was hoping that would explain this a little better.

Florida (at #9 as an opponent, highest among teams with losses) is the highest-rated win this season, so that’s why I was thinking it might be enough for Alabama, but Penn St. ranks 27th as an opponent, lower than West Virginia, whom LSU beat.  Arkansas is 28th, and Mississippi St. is 29th, so Alabama’s best second-best and third-best opponents cancel out LSU’s best and second-best opponents.  But then LSU catches up.  North Carolina is a respectable 43rd, and Tennessee is 67th.  Duke and San Jose St., neither of whom have an FBS (I-A) win, are #109 and #111, respectively.  LSU’s lowest-rated win is Vandy, who is #74 as an opponent.  If Penn St. beats Iowa, Duke beats Maryland (close game), and San Jose St. beats UC-Davis (close game), along with maybe Miss. St. and UNC losing, that might have all combined to put Alabama ahead of LSU.  The distance between Alabama and Oklahoma is not very much, so those three Alabama opponents’ winning might have been enough even if Oklahoma stays the same.  For instance, I accidentally had North Carolina 3-1 (I double-checked every team after noticing that error), and that would have given LSU an extra .07 in the ratings.  Oklahoma and Alabama are less than .04 apart, and LSU is less than .02 ahead of Oklahoma.

Maybe this year Vandy finishes 1-11 and Penn St. finishes 11-2, but it goes by where everyone is right now, including opponents and opponents’ opponents and even to some extent their opponents (because I don’t think you can fully rate the opponents without this information).  Also, it doesn’t look ahead and say that LSU stands to gain fewer points because they have two easy in-state opponents later in the year. So that’s another factor.  If LSU plays WVU and UNC later and those weaker opponents earlier, LSU wouldn’t be in the conversation.  If combined with that, BYU is better this year and beats Air Force and Utah St. (coincidentally two of Oklahoma’s opponents), Alabama is probably #1 right now.   It’s going to take a few more games for the ratings to reflect the teams’ acoomplishments more accurately without it being a function of scheduling quirks and opponents who start out differently from how they will end up later.