Michigan State actually has 5 quad 2 wins (they had 3 as of Friday) A couple teams must have played their way up into a quad 2 win for them.
So they had 8 quad 1/2 wins not the 6 I thought they had so that was not as few as I thought.
Their non conf SOS was awful though that might been why so low.
This exemplifies why the tier system is a poor choice for the committee to use. Whats better, beating #1 or #30? The tier system doesn't differentiate.
- First, scrap using only the RPI for team comparisons. ITs not a bad system to use, but combine it with the other metrics like SOR, BPI, KenPom, etc to come up with an average computer ranking for each team.
- Second, all of these ranking systems already compensate for location of the games. RPI for example gives you a 1.4 credit for road wins and a .6 credit for home wins, so no need to make your tiers up by giving extra credit because the rnakings have already raised their ranking accordingly.
- Third, get rid of the tiers all together and use an average ranking of best 5 wins, best 10 wins and overall record. For example, if Team A has a best 5 wins of 2,7,8,13,20 and team B has a best 5 wins of 12,15,18,25,30, Team A would have an average ranking of 10 and team B would have an average ranking of 20. Isn't that metric better for comparison than saying both have 5 Tier 1 wins? IF you think winning on the road or neutral needs to be rewarded more than the computer models already provide, then take 25% off a teams ranking for a neutral win and 50% off for a road win (ie, beating #40 at a neutral site would mean using 30 when computing that wins value or 20 if it were a true road win).