Sunday, October 23, 2011

College Football Week 8 BCS Observations - Oklahoma State, Stanford, Kansas State, Houston

The latest BCS standings are out after week 8's games, and to no ones surprise LSU and Alabama are at the top.  And I have no complaint with that as my computer has them in a virtual tie at #1 and #2, just in the other order.  And see this for a discussion of the details of the two teams ratings.  But there are still some interesting observations one can make from the latest BCS and specifically the computer rankings that go into them.

Let me preface my comments by saying that I disagree with the BCS mandating that the computers not use margin of victory (MOV) at all.  I believe that a proper system can use MOV to accurately rate a team and not punish too much for playing a weak opponent or two but also not reward running up the score.  The BCS differs apparently and we see some screwy rankings as a result.

The first observation is that it is quite surprising that all but one computer has Oklahoma State #1.  My computer does have them #3, but they are over 5 ratings points back of LSU and Alabama.  As can be seen on their performance chart, they have played a reasonably tough schedule, but the computers should not be factoring in their big MOV in several of their games and only 4 opponents have winning records and one of those is Louisiana-Lafayette which my computer has #100.  The good news is they have to run the gauntlet of Baylor, K-State, Texas Tech, and finish with Oklahoma, so we'll know soon if they deserve what the BCS computers are giving them credit for.

The BCS has Stanford #6 which is the same as my computer, but several computers seem way off in where they have them ranked.  Yes, they have not played the tough part of their schedule yet, but #14 and #21?  This is where not allowing MOV shows its shortcoming as it handicaps a computer to not be able to estimate that a team is very good even though they have played a weak schedule to date.

Kansas State on the other hand, my #14 and BCS #8, is also undefeated against only a slightly tougher schedule, but the BCS computers really like them with an average ranking of 5 after throwing the low of #9 out.  That seems too high, but again they will be tested soon.  Will I be eating my words and admitting the BCS computers had it right or will my #14 be more accurate in a month?

Houston presents a unique set of challenges in determining where they should be.  They are undefeated but against very weak opponents and my computer has them #36.  And they've had a number of close games, 1 point win over Louisiana Tech, 4 point win over UCLA, 7 point win over UTEP, only UTEP at 4-3 with a winning record of those 3, but again the computers not being able to use MOV handicaps them as they can't reflect them struggling in these games.  The result is 2 computers have them in the top-10 at #9, another at #14, and a fourth at #16.  Does ranking Houston in the top-10 pass the smell test?

Let me know what you think, leave a comment.