Welcome!

By registering with us, you'll be able to discuss, share and private message with other members of our community.

SignUp Now!

Pomeroy

There are four teams with double-figure losses in Pomeroy's Top 50 now: Wake Forest at #30, Clemson at #39, Syracuse at #49, and UGA at #50.
 
After the win at St. Mary's, Gonzaga has the 3rd highest efficiency margin in the history of Kenpom. Kentucky 2015 leads everyone by a wide margin, then Kansas 2008 is 0.01 (the smallest possible margin) ahead of Gonzaga now.

Gonzaga is going to be my pick in any bracket contest/office pool. I don't think they will have more than a 20% chance to win it all, but they present great value as a historically dominant team that relatively few people are going to pick based on perception. My 2nd value pick would be Duke from a 3 or 4 seed. Lol.
 
SeanMayTriedToEatMe said:
The Clemson win was nice, despite the margin, in that it bumped Duke's projected ACC record up to 11-7. I think it was still 10-8 after the UNC game. That projection has Duke finishing as the 5-seed in the ACCT, ahead of ND due to the H2H tie breaker, but ND might get their projection bumped up with a win over FSU today. Had Duke beaten NC State at home, they'd be projected for a 3-way tie for the 3-seed.
Click to expand...
Duke was projected at 11-7 after the UNC game. That hasn't changed. According to kenpom odds, Duke has a 1% chance to win out. Duke almost controls its own destiny for the ACC #1 seed other than needing Louisville to lose one more game. GO DUKE!!!
 
physicsfactor said:
SeanMayTriedToEatMe said:
The Clemson win was nice, despite the margin, in that it bumped Duke's projected ACC record up to 11-7. I think it was still 10-8 after the UNC game. That projection has Duke finishing as the 5-seed in the ACCT, ahead of ND due to the H2H tie breaker, but ND might get their projection bumped up with a win over FSU today. Had Duke beaten NC State at home, they'd be projected for a 3-way tie for the 3-seed.
Click to expand...
Duke was projected at 11-7 after the UNC game. That hasn't changed. According to kenpom odds, Duke has a 1% chance to win out. Duke almost controls its own destiny for the ACC #1 seed other than needing Louisville to lose one more game. GO DUKE!!!
Click to expand...

We're going to be a victim of the schedule down the stretch. We have to make up ground or gain ground on the teams we're tied with, but there's also the issue of the teams right behind us. It was cool that ND beat FSU but ND is now 8-5 with the following games left: @BC, @State, GT, BC, @ Louisville. At least we have the H2H tiebreak over them.
 
So we've played 5 of the current top 10 KenPom teams. We have victories over #3 UVA, #6 Florida, and #10 UNC; we have losses to #5 Louisville and #9 Kansas.
 
I haven't paid as much attention to kenpom this year, does it feel like records/rankings match up with it more or less than previous years. It feels like less, but maybe I'm just making that up
 
UVA had a top 10 offense at some point during the ACC season, I believe. Currently 51st.
 
Basically, Pomeroy should no longer be considered predictive. It is lagging or coincident at best.
 
I just spent a while compiling every Duke player's OBPM and DBPM from each season as far back as those stats go on sports-reference.com (2011).

The good news is that our players do indeed get better over time (i.e., our seniors on average have higher OBPM and DBPM than our juniors, our juniors on average have higher OBPM and DBPM than our sophomores, and our sophomores on average have higher OBPM and DBPM than our freshmen).

The bad news is that my goal was to discover some discrepancy between the improvement over a career on offense vs. the improvement over a career on defense, and I found none. Offense improves pretty much at the same rate as defense as players go through the Duke system.

What might be more enlightening is if someone went through each season since kenpom begins (2002) and calculated a weighted average minutes distribution. Add up the minutes to all seniors, multiply by 4. Add up the minutes to all juniors, multiply by 3. Add up the minutes to all sophomores, multiply by 2. Add up the minutes to all freshmen. Add up the total of these numbers and assign that number to that season. Chart the correlation between that weighted average minutes distribution and adjusted offensive efficiency, along with adjusted defensive efficiency.

I think the hypothesis would be that a higher number correlates to a better defensive efficiency significantly more strongly that it does to a better offensive efficiency. In Duke's case, I don't think it's possible to find a stronger correlation to offense, since Duke's offense has remained so consistently good, with only 1-2 down seasons. With defense, we have plenty of down seasons to strengthen the correlation.

I'll do this at some point.
 
Do you know if the "Experience" category on Kenpom gives weight to minutes actually played? If not, it prob should (or else it's kinda worthless). And if it does, cant you just use that?
 
That's a good question. Really, he should be separately measuring minutes played and average age.
 
Actually that is exactly what kenpom does (weighs experience by minutes played, discounts anyone under 10% of minutes), so all we need to do is chart the correlations.

Kenpom also has a "minutes continuity" stat, and he charted the correlations already:
http://kenpom.com/blog/measuring-continuity/

Not surprisingly, teams with more continuity tend to perform better, with the effect being stronger on offense.
Click to expand...

Not what we're looking for as an explanation for Duke's defense since 2010, when considering the offense has remained pretty constant.
 
The experience correlation for Duke is all over the place, with not enough data points (experience only goes back to 2007 on kenpom).

2007, 2008 and 2015 were absurd outliers for defensive efficiency while having inexperienced rotations. Josh McRoberts and DeMarcus Nelson must have been monsters. 2015 we know was a matter of catching fire defensively at the right time.

Somewhere out there is a good way to measure what we want to measure here. It would involve recruiting rankings, experience, defense and offense. But I can't think of a good way to make this happen in a useful way.
 
Recruiting rankings are all over the place. I don't know how you could make them work. Bolden is my poster child for this.

He was ranked:

34 - 247
20 - ESPN
16 - Rivals
12 - Scout (probably why they went out of business)
USA Today comes up with a composite of 19th

Just averaging the numbers makes no sense, not does throwing out highest and losest and averaging the middle. In Bolden's case, you could probably just add all the numbers up, multiply by 2, and come up with a reasonable ranking.
 
Yeah, the goal is basically to determine the perfect way to construct a roster and see what K has been doing wrong, if anything. I don't think we can do that with the limited time on our hands as normal working folks. Someone who gets paid for that stuff or who can use it as part of a college/grad school thesis would have to be the one to figure it out.
 
I don't think you just used minutes played though. I think time spent in the program still matters, for one thing a couple of years of strength and conditioning (and age related physical maturing) and practice time alone should make someone a much better defensive player than someone just coming into the program even if they both have no actual game minutes.
 
You must log in or register to reply here.

Chat users

  • No one is chatting at the moment.

Chat rooms

  • General chit-chat 0

Forum statistics

Threads
1,065
Messages
423,978
Members
624
Latest member
Bluegrass Blue Devil
Back
Top Bottom