http://www.washingtonpost.com/sports/nca...ory_1.html
The sentiment of some rating gurus is that the RPI should be used as nothing more than a rough tool. Joel Sokol, an engineering professor at Georgia Tech whose ratings system (LRMC) is used by the selection committee, said the RPI is “not even close to the best indicator of a team’s strength.”..........................“I don’t like the RPI,” Pomeroy said. “It has no analytical value whatsoever.”.......................No one painted a more visual image of the RPI as a relic than Scott Van Pelt, the ESPN personality who, in calling it the worst metric in sports during a seven-minute radio rant, likened the reliance on the RPI to a man “walking around with a big Walkman on his hip the size of his toaster, who is flipping over his cassette tape, who wants to run home to program his VCR on his standard-definition television.”
I think they should start with overall record, and then look at quality wins. I also think you're last 10 to 15 games are more important than the your first 15. How you're doing finishing the season can say alot about how you'll compete in the Big Dance.
UC played a weak OOC schedule, but won 12 games in one of the top 2 conferences in the country, along with some quality home AND road wins in that conference schedule. Their RPI and SOS is tainted by that weak OOC schedule. But it argues for looking deeper as their Big East resume' is outstanding.
I think overall record is not that important. I think they should use some blend of things like KenPom and Sagerin (not RPI) and then use those similarly to how they use RPI. Look at top 100 wins, top 50 wins. See how many wins you have over really bad teams (sub 250 wins just don't tell you anything and should not be looked at, which is why I don't worry about overall record). Look at who you lost to and the type of teams you lose to.
I actually like the way the committee evaluates teams. I think the way a described is pretty much how they do it, except that they use the RPI which is just not as good a lens to evaluate through. Use the better computer rankings as the lens, and continue to look for the same types of characteristics and I think the committee does an even better job (and honestly, in general I think the committee does a pretty good job. I am happy with the way the field looks most years).
I don't like the RPI. I look at a team like Southern Miss that was in the top 20, but really had no real signature win, but they do have lots of top 100 wins.
I understand not wanting to encourage running up the score, but removing the margin of victory is just stupid. Maybe you cap it or don't weigh it much. But a 1 point overtime win or loss should be compared with a 25 point win or loss differently (Not much, just some).
A one point overtime loss to Syracuse should weigh more (Or at least be worth something) than a 1 point overtime win against NC Asheville. Not to mention a road win is weighed too darn much. You often get more credit for beating a team ranked in the 100s on the road than beating a top 25 team at home.
Plus with the RPI there are plenty of schools admittedly charging a ton more because playing them is expected to help a teams RPI. It helps both the haves and the better RPI schools by helping inflate both teams RPI. Then the have nots have an even tougher time of improving their RPI as do those that can't afford to pay for the better RPI teams since both teams RPIs are lower and bring boths teams down even further.
I would like to see the old rpi, kenpom, and sagarin used to create computer ranking average or range.
The RPI has trended towards garbage since they updated the calculations. Coaches knew this years ago. Rest of the world is finally catching on.
They should average the Sagarin, KenPom, ESPN BPI and the Massey Ratings...
(03-05-2012 08:43 AM)bearcatmark Wrote: [ -> ]I think overall record is not that important. I think they should use some blend of things like KenPom and Sagerin (not RPI) and then use those similarly to how they use RPI. Look at top 100 wins, top 50 wins. See how many wins you have over really bad teams (sub 250 wins just don't tell you anything and should not be looked at, which is why I don't worry about overall record). Look at who you lost to and the type of teams you lose to.
I actually like the way the committee evaluates teams. I think the way a described is pretty much how they do it, except that they use the RPI which is just not as good a lens to evaluate through. Use the better computer rankings as the lens, and continue to look for the same types of characteristics and I think the committee does an even better job (and honestly, in general I think the committee does a pretty good job. I am happy with the way the field looks most years).
Well, basically I agree.
But I look at overall record first and then analyze if they were bunny wins or against quality teams. Sure, many teams pad their 25 to 28 wins with an easy schedule or easy conference (which they can't necessarily help)......I agree there.
And there are also teams that accrue only 18 wins that might have played a killer schedule, also deserving.
But the point in starting with overall record is that you don't bother with the 15 wins or less teams. List every team that won 16 or 17+ games and then analyze their schedule.