Started By
Message
NCAA Considering Composite Formula to Select & Seed 2018 NCAA Basketball Tourney
Posted on 2/16/17 at 9:25 am
Posted on 2/16/17 at 9:25 am
LINK
Interesting idea. Last year's committee chair said that they used RPI and KenPom most heavily fwiw.
quote:
So the NCAA invited the game’s top mathletes to its headquarters Friday to discuss the possibility — and obstacles — involved in developing such a “composite” rating, perhaps in time for use in selecting and seeding the 2018 tournament.
In addition to Pomeroy, the NCAA welcomed Jeff Sagarin of USA Today; Michigan State administrator Kevin Pauga, who conceived the KPI ratings; Ben Alamar of ESPN, which conceived the BPI ratings several years back; bracket analyst Jerry Palm of CBS Sports; and Andy Glockner, author of the basketball analytics book “Chasing Perfection.”
Interesting idea. Last year's committee chair said that they used RPI and KenPom most heavily fwiw.
4
Posted on 2/16/17 at 9:27 am to volfan30
I'm all for taking the human bias out of a selection process
This post was edited on 2/16/17 at 9:28 am
Posted on 2/16/17 at 9:28 am to volfan30
As a math guy myself I like KenPom's system and my own NBA system is based on the same exact principles
The issue is that there are still biases in it. Any and every system will have bias. It's impossible not to have bias.
The issue is that there are still biases in it. Any and every system will have bias. It's impossible not to have bias.
Posted on 2/16/17 at 9:29 am to thunderbird1100
quote:
I'm all for taking the human bias out of a selection process
Um, you really don't understand computers do you... something about input and output might ring a bell.
This post was edited on 2/16/17 at 9:30 am
Posted on 2/16/17 at 9:31 am to goldennugget
I like KenPom but I don't think the NCAA Tournament should be selected based on KenPom.
The NCAA tournament should first and foremost select teams based on what they did during the year. KenPom measures the strength and how "good" a team is. It's great for analyzing matchups and determining who might win a game, but it doesn't pay attention to wins and losses. A team could blow out 3 Top 20 teams and lose close, unlucky games to 5 100+ teams and KenPom would have them very high. However, winning matters.
If you are down to two teams with very similar records, KenPom should come into play. But I wouldn't want the tournament picked that way.
The NCAA tournament should first and foremost select teams based on what they did during the year. KenPom measures the strength and how "good" a team is. It's great for analyzing matchups and determining who might win a game, but it doesn't pay attention to wins and losses. A team could blow out 3 Top 20 teams and lose close, unlucky games to 5 100+ teams and KenPom would have them very high. However, winning matters.
If you are down to two teams with very similar records, KenPom should come into play. But I wouldn't want the tournament picked that way.
Posted on 2/16/17 at 9:36 am to TheRaid
quote:
Um, you really don't understand computers do you... something about input and output might ring a bell.
I understand them just fine, in fact been building them for well over a decade. Of course a program can be as biased as the programmer makes it, the point of this is probably to get a set of metrics though that take the current ever changing human bias out of the selection process so teams know exactly what to expect and what they need to do on their schedules to make the tournament.
The biggest problem with the committee is the selection criteria changes not only year to year with no one knowing about it, but on top of that changes from team to team they select. There's no continuity. There's teams every year who feel slighted by not being chosen because of some "reason" the committee makes up just for them. Like a mid major who is 26-6, RPI of say 39, gets left out because "not enough wins over Top 50 teams", say they have only 2, but other teams made the tournament who had only 2 (or less) RPI Top 50 wins.
When we're talking 68 spots here, I have no issue with a computer formula choosing a little more than half that field after conference champs. Makes a lot of sense.
This post was edited on 2/16/17 at 9:39 am
Posted on 2/16/17 at 9:40 am to thunderbird1100
quote:
When we're talking 68 spots here, I have no issue with a computer formula choosing a little more than half that field after conference champs. Makes a lot of sense.
Agree. You can very easily create an agreed upon set of criteria for how much things like road wins, Top 50 RPI wins, KenPom rating, Sub 100 losses, conference wins adjusted for conference strength, etc count and come up with a total formula that spits out an end number that ranks all the non-automatic qualifiers. Most importantly, it will be consistent year over year, as well as completely transparent.
I'm actually a bit surprised nobody has created something like that already.
This post was edited on 2/16/17 at 9:44 am
Posted on 2/16/17 at 9:54 am to goldennugget
The problem with KenPom is that it isn't made for NCAA Tournament selection
It is made for individual game predictions and he even says as much
It is made for individual game predictions and he even says as much
Posted on 2/16/17 at 9:56 am to GenesChin
quote:
It is made for individual game predictions and he even says as much
Yep. It's supposed to tell you how good a team is. It is not made to reflect a team's resume.
Like I said, if you get down to it and both resumes are even, then I'm all for brining in KenPom to "break the tie" so to speak.
Posted on 2/16/17 at 9:59 am to Triple Daves
I think it was Sagarin who said margin of victory needs to be scaled with say every W is assumed by some arbitrary number like 50pt + actual margin. By doing this you give credit for winning big but reduce the impact since BBall margin is a fickle thing
Right now a ranking like KenPom penalizes teams for putting walkons in and/or not running up the score
Right now a ranking like KenPom penalizes teams for putting walkons in and/or not running up the score
This post was edited on 2/16/17 at 10:07 am
Posted on 2/16/17 at 10:14 am to volfan30
Anything the NCAA does is terrible so this may not be a bad idea.
Page 1 of 1
Popular
Back to top
Follow SECRant for SEC Football News