PDA

View Full Version : NCAA Selection Changes for 2018 - Good or Bad for GU and Small Conference Teams?



75Zag
03-13-2017, 06:02 AM
Various sources (Palm, etc.) report that the NCAA will vote this summer to move away from using the RPI toward some sort of Tournament selection/seeding process that will rely more heavily on yet-to-be determined metrics, perhaps along the lines of KenPom and Sagarin's systems, but adjusted to reduce the importance of "margin of victory". As long as GU remains in the WCC, GU is going to have major problems with SOS compared to Blue-Blood teams. I think GU does well under KenPom, but I wonder if much of GU's success under KenPom is due to the margin of victory metric, and if the importance of that metric is reduced or eliminated, I wonder how well GU would do under that or any similar system.
Is this proposal a clever PR move by the NCAA which will actually give SOS (aka Power Conference Teams) more weight in NCAA Tournament selection/seeding, or does it actually move the NCAA closer to some sort of hypothetically perfect system where no great teams are left out regardless of conference and no "dud" teams from Power Conferences are invited?

See you in SLC - Go Bulldogs!

thegloriousgoateeofKP
03-13-2017, 08:41 AM
Wish they wouldn't reduce the importance of margin of victory, but as long as the new system properly adjusts for the difference between home and road, it'll be way, way, WAY better for mid-majors than the RPI is.

CDC84
03-13-2017, 08:43 AM
They have to reduce margin of victory to some degree or you will see Duke beating low major teams by 75.

thegloriousgoateeofKP
03-13-2017, 08:49 AM
They have to reduce margin of victory to some degree or you will see Duke beating low major teams by 75.

Any team that Duke can beat by 75 will, ideally, be ranked very low in their new system...which (again, ideally) will properly adjust for strength of schedule.

amaronizag
03-13-2017, 09:00 AM
Ken Pomeroy doesn't really use the margin of victory as a metric. For example, his system compares the number of points GU scores per possession with the number of points our opponents score per possession against us. He compares all teams in that way, and that is one of the ways the committee will compare teams next year. Since that is a tempo free statistic, it doesn't really compare scores, the AdjEM allows people to compare the efficiency of each team on offense and defense. You can look at more detailed metrics for each team and see how their scoring and other stats compare to the mean or other teams, but scores and winning margins isn't really analyzed by Pomeroy.

Hoopaholic
03-13-2017, 10:29 AM
I personally believe they should set aside 2 seats in each of the first 4 lines and then 3 seats in lines 5-12 dedicated SOLELY to the conference champions.

Conferences can be put on an S curve and placed into the appropriate lines based upon the toughness of their conference.

This would force the middle of the pack power 5 to compete across the board against conference winners and would reward the conference winners for their body of work in a dedicated fashion by power curve.

soccerdud
03-13-2017, 12:18 PM
Ken Pomeroy doesn't really use the margin of victory as a metric. For example, his system compares the number of points GU scores per possession with the number of points our opponents score per possession against us. He compares all teams in that way, and that is one of the ways the committee will compare teams next year. Since that is a tempo free statistic, it doesn't really compare scores, the AdjEM allows people to compare the efficiency of each team on offense and defense. You can look at more detailed metrics for each team and see how their scoring and other stats compare to the mean or other teams, but scores and winning margins isn't really analyzed by Pomeroy.

you are correct about him attempting to control for tempo and opponent-- but in all cases the relationship between margin-of-victory and the composite (o+d) efficiency evaluation is such that the latter increases as the former does. you hit one more shot or get one more stop, and your rating improves. always. this might not be technically linear in all cases, but it is a strong positive correlation. put another way, comparing "the number of points GU scores per possession with the number of points our opponents score per possession against us", multiplied by the possessions in a game IS our margin of victory.

normalizing for opponent and pace does not change that. win by more, your rating goes up, regardless of how you label the metrics. this incentivizes winning by more.

personally, this bothers me not at all. but the idea that "kenpom doesn't explicitly factor margin of victory into his rankings" is baloney. his rankings are effectively NORMALIZED margin of victory/defeat.

CDC84
03-13-2017, 12:29 PM
And what about a team like Kansas who has a number one seed but who seems to have a margin of victory of 5 against Big 12 opponents who are highly rated? That team has won so many tough, close games. I don't know. I have mixed feelings about margin of victory because I still feel accomplishment matters the most. At the end of the day, it takes 6 one point wins to win the NCAA title. I know, pigs will fly before there is a champ that wins all their NCAA tourney games by one point :) But there have been some final four teams who didn't exactly blow away the competition to get to the final 4.....even in the first round. The Arkansas team that lost to UCLA in the title game in the 90's comes to mind.

maynard g krebs
03-13-2017, 02:49 PM
And what about a team like Kansas who has a number one seed but who seems to have a margin of victory of 5 against Big 12 opponents who are highly rated? That team has won so many tough, close games.

Well, when you're allowed to tuck the ball under your arm like a running back just across mid court and carry it to the rim, it tends to help you win close games.

CDC84
03-13-2017, 03:45 PM
Well, when you're allowed to tuck the ball under your arm like a running back just across mid court and carry it to the rim, it tends to help you win close games.

Any new statistical model for measuring teams must include bad officiating.