Chaos Rankings

Monday, October 16, 2006

Week 7 Top 24

Previous Rank
Score Change
11.0Ohio State10.0
30.968291Boise State3-0.014432
40.939569West Virginia6-0.003322
160.502767Texas A&M180.009702
200.475947Notre Dame17-0.022969
220.460103Georgia Tech20-0.029047
240.449608Boston College370.036103


  • Wow, your poll actually makes sense concerning Florida and Auburn! Florida loses to a good Auburn team, yet Auburn loses to a decent-at-best Arkansas team. For some reason, the other polls thought Florida was nine spots better than Auburn last week, but this week Florida is barely a top-ten team. I still don't understand how recent losses matter more than earlier-season losses.

    By Blogger m├Ątt, at 10/18/2006 7:09 AM  

  • Which is why conference championship games can ruin a season. Say you have a Florida/Auburn rematch for the SEC title. Auburn by that point gets up to #3 in the rankings after Michigan loses to Ohio State. USC takes a late-season nose dive, but is still clinging to #2. An 2nd win over Florida would lift the Tigers into the #2 spot, but they lose to UF instead. Computers help their cause with a strong SOS, but the pollsters doom them by dropping them 2 spots because they've now got 2 losses. The best thign you can do is lose early, cause then people attribute it to "early season jitters" or whatever. No one's talking about GT's loss to Notre Dame anymore. Lose your first game to drop just off the radar, then run the table, and campaign like crazy. But god help you if you win out and then lose your conference championship game. Oh, you were undefeated and headed to a BCS championship, but you lost your conference title, so now you're not even guaranteed a BCS game at all, and if the numbers work out right, maybe you get an at-large bid.

    Fortunately, computers seem less prone to that sort of shenanigans. A loss is a loss, regardless of when it occurs. In my system, a loss always counts for the same thing. The only things that's different early vs late is your schedule strength, which is recomputed every week based on current performances. So if a team you beat early is on a 4-0 tear, you get a boost in your SOS, but then they turn into Michigan State and suddenly 4 weeks later, your SOS has tanked because a team that looked like they could hang with anyone rolled over and died. Rather than take a "snapshot" SOS of how good a team was when you played them I take a total season look at how good a team you played has been and use that in the SOS calculation. For example, Notre Dame was a preseason #2 when they beat GT, but now they're hovering around 10-15. So rather than use that #2 to say how good GTs SOS is, I would use the #13 or whatever. Then if Notre Dame takes a nose dive in the second half of the season and end up #30, I would use a #30 rating to calculate the SOS at that point in time.

    By Blogger Adam, at 10/18/2006 2:31 PM  

Post a Comment

<< Home