Louis Gossett Jr
Well-known member
- Joined
- Sep 4, 2012
- Messages
- 12,971
- Reaction score
- 6,886
Why can’t it be a weighted exponential decay?
Tie goes the runner, so whoever scores the most points obviously.I do not think the metrics are the end-all be all (despite my defense of them on here). They are flawed and made by imperfect beings.
That being said, for those who think they are "shit and don't work", how would you go about selecting the NCAA Tournament field?
What would you use, what wouldn't you use? Why/why not?
It's easy to criticize what is being used/done, but it's more difficult to provide your own thoughts on how YOU personally would select the field on Selection Sunday.
For instance, how would you compare Boise State's resume to West Virginia's resume if they were the final two teams up for one spot?
Yep sure hasI think I have heard Kenny Pom himself say the committee should look at resumes (wins and losses) more than his rankings.
Yeah I’m much more interested in NCAA selection based on what a team has accomplished vs selection based on predictive elements.Another question is do folks think metrics suck and don't give any real information at all, or just that the NCAA shouldn't be relying so heavily on predictive elements when selecting teams and instead should lean more on resume-based elements? Because I think these are two totally different things. The latter I find to be an interesting conversation, the former makes me wonder if kids these days take statistics courses anymore
Yeah and from the conversations in how they put the NET together (got all these analytics guys to come in and sit down a few times with the NCAA folks), I suspect this was the aim for the NET. Again, I find this conversation far more fruitful than "metrics suck because they say Florida Atlantic is a top 35 team."I would not select the field by going down the kenpom rankings or any other similar ranking system. At the same time I would find it difficult to select one team over another if the various metrics consistently had that team 30-40 spots below the other, barring unusual circumstances.
You don't see a problem with that? At a certain point we get a little too cute with selections and ignore this important part of atheltics.I also would not select teams solely by KenPom, as his model does not care at all if you win or lose games.
Well his model is predictive not resume based so no, if you're not picking the tournament at large teams solely by metric rankings (which the committee isn't), then I don't see the issue since winning close games hasn't proven to be predictive over the tens of thousands of data points that we have since advanced metrics have gone more live/mainstream.You don't see a problem with that? At a certain point we get a little too cute with selections and ignore this important part of atheltics.
Not at all, as his model is for predictive purposes, not picking a field of 68.You don't see a problem with that? At a certain point we get a little too cute with selections and ignore this important part of atheltics.
Another great point. Punishing a team for a bad OOC if they have solid adjusted metrics is definitely double counting.I am just bitter because last year we had the wins and the predictives. Yes we had a terrible non conference SOS, but that was baked into the metrics already.
It is almost certainly us and 13, but we did go to 20 conference games last year.What is the most number of ACC wins/highest ACC Win% a team has had and not made the NCAA tourney (since expansion)
Yeah and from the conversations in how they put the NET together (got all these analytics guys to come in and sit down a few times with the NCAA folks), I suspect this was the aim for the NET. Again, I find this conversation far more fruitful than "metrics suck because they say Florida Atlantic is a top 35 team."
Since 2005:What is the most number of ACC wins/highest ACC Win% a team has had and not made the NCAA tourney (since expansion)
Sure so folks don't want metrics to be the end all be all and want the committee to use other factors in addition to the metrics to make decisions. That's what they do. Wake missed the tourney at 42 last year and Rutgers got in at 80. The committee is able to parse through the NETI think it's the same conversation. I can't speak for everyone bringing up the examples, but I'm pretty sure nobody here cares if Texas Tech would be favored over WF at a neutral site. They do care that Texas Tech was ahead in NET until last night (and is still ahead in KP), since NET Is the official NCAA selection analytics tool, and since TT has zero wins over the top 200 with the exception of home games vs. Eastern Washington and Louisiana Tech