Our favorite former meteorologist talks quality wins...

[kso_FAN]

http://kenpom.com/blog/index.php/weblog/entry/fairness_in_quality_wins

His argument (IMHO a very good one) is this:

The problem is that road win over #51 is significantly more difficult than a home win over #50. We can all agree on that, right? And yet one counts as quality and the other doesn’t.

And he goes on to say this.

     
against…
         Home   Neutral  Road
Win%    Opp Rk  Opp Rk  Opp Rk
59        20      50      73
76        57     100     177

You could make the baseline team whatever you want, but it wouldn’t change the calibration much. Instead of focusing on games against the top 50, the NCAA could call it games against quality opponents or Tier I opponents, or something catchy that would get a hashtag. Whatever you call it, that group would include home games against the top 20, neutral games against the top 50, and road games against the top 75. I can’t imagine the code change would take more than an hour for the NCAA to implement. Boom, now Oregon has a quality win it deserves, and teams picking off the 47th ranked team at home don’t. (Using my ratings, anyway.)

[mocat]

so KU should be punished for losing to SDSU at home eh?

[Cartierfor3]

so KU should be punished for losing to SDSU at home eh?

x infinity

[michigancat]

His method of "tier 1 wins" as opposed to quality wins basically has the same problem in that a road win over the 70th ranked team really isn't better than a win over 71st.

I honestly think the committee would be better off using Kenny's algorithms instead of a committee at all.   

[sys]

yeah, binning teams into categories is stupid.  any method that relies on that is intentionally flawed.

[Kat Kid]

Instead of binning teams is there a suggestion that some sort of average should be used or should each team have their wins and losses simply rank ordered? 

There needs to be a simple way to facilitate comparisons and discussion.  I think a rank order would do it.

[kso_FAN]

a) I think it is fun/good to have discussions about the criteria the committee uses to select teams. RPI has just become the popular target with the rise of more advanced stats.

b) Honestly, the committee does a pretty good job selecting 68 teams (or whatever it is now) and seeding for interesting match-ups and a fun to watch tournament.

[mocat]

Talking heads always say "team x has xx wins over top 50 teams, as well as road wins over y and z"

[pissclams]

There needs to be a simple way to facilitate comparisons and discussion.  I think a rank order would do it.

there is a simple way to facilitate comparos.  each team can be assigned a colored belt.  some teams are browns.  some are yellows.  some are oranges.  the team then wears the belts as they play so their ranking can quickly be identified by viewers who will assess whether or not the team's belt designation is appropriate.  after watching the team play, viewers can press a colored button which corresponds to the color of the belt which they feel the team should be wearing.   /thread #simple #belts