Dork Poll tracking (Pomeroy-Sagarin)

[JasonEvans]

(note-- I meant to post this last night, but forgot to hit the submit key-- duuh!)

By the way, if anyone cares, Marquette was the #33 team in Pomeroy's rankings before tonight. Ken had forecast a final score of 80-67 (which would have been darn close to right had Duke not gaaacked up a fair amount of the lead in the final 30 seconds or so).

Just so we can see how it changes-- Marquette's overall rating was .8929. Its offensive rating was 112.0 (#27) and its defense was 93.1 (#49).

Duke was the #1 team with an overall rating of .9844. Duke's offense was 121.7 (#1) and the D was 84.8 (also #1).

--Jason "geek much?" Evans

[77devil]

(note-- I meant to post this last night, but forgot to hit the submit key-- duuh!)

By the way, if anyone cares, Marquette was the #33 team in Pomeroy's rankings before tonight. Ken had forecast a final score of 80-67 (which would have been darn close to right had Duke not gaaacked up a fair amount of the lead in the final 30 seconds or so).

Just so we can see how it changes-- Marquette's overall rating was .8929. Its offensive rating was 112.0 (#27) and its defense was 93.1 (#49).

Duke was the #1 team with an overall rating of .9844. Duke's offense was 121.7 (#1) and the D was 84.8 (also #1).

--Jason "geek much?" Evans

Numbers have been updated and Marquette moved down a notch, although its rating edged up to .9009. Duke lost a little ground in all 3 ratings, but remained #1 in them.

I am surprised that Marquette's O and Duke's D #s didn't move a bit more given the Golden Eagles offensive efficiency in yesterday's game.

[COYS]

Numbers have been updated and Marquette moved down a notch, although its rating edged up to .9009. Duke lost a little ground in all 3 ratings, but remained #1 in them.

I am surprised that Marquette's O and Duke's D #s didn't move a bit more given the Golden Eagles offensive efficiency in yesterday's game.

Marquette didn't manage a point per possession against us. While that is still significantly more than our other opponents, it is not particularly impressive. Our D allowed a little bit more per possession than it probably should have, but the 19 turnovers were the real culprits in hurting our Pomeroy stats. That killed our efficiency. Actually, we were pretty efficient when we were actually able to get a shot off without coughing it up (I'll take 55% shooting any night!). I'll say more in the post game thread, but I'm honestly not too worried about the turnovers at this point and would expect to see them come back down as the team gets accustomed to recognizing when it should play fast and when it should slow it down a bit.

[gam7]

Do the offensive and defensive efficiency ratings take into account strength of schedule? That is, would a given set of offensive and defensive statistics be weighted more heavily against a good team than it would against a bad team (e.g., if Duke has an eFG% of 50% against Colgate, it's not as impressive as it is against K-State). Or, do you have to look at strength of schedule separately, as in "Team A is ranked #5 in offensive efficiency, but they've done it against a #1 ranked strength of schedule."

Also, what does the "pyth" measure actually mean in layman's terms?

[loran16]

Do the offensive and defensive efficiency ratings take into account strength of schedule? That is, would a given set of offensive and defensive statistics be weighted more heavily against a good team than it would against a bad team (e.g., if Duke has an eFG% of 50% against Colgate, it's not as impressive as it is against K-State). Or, do you have to look at strength of schedule separately, as in "Team A is ranked #5 in offensive efficiency, but they've done it against a #1 ranked strength of schedule."

Also, what does the "pyth" measure actually mean in layman's terms?

Yes, the ADJUSTED O and D efficiency ratings take into account strength of schedule. That's a good reason why Duke was ranked #1 in offensive efficiency last year...as our our opponent's had the best D in the NCAA (the ACC was well, kind of good defensively). Pomeroy used to have listed the raw offensive and defensive efficiency numbers, which don't take into account opponent's, but I can't find it up there on the new version of the site.

Pyth (Short for Pythagorean, like the triangle. Long story.) is the result of a formula which takes the offensive and defensive efficiency to find an expected winning percentage over a regular season.

[gam7]

Yes, the ADJUSTED O and D efficiency ratings take into account strength of schedule. That's a good reason why Duke was ranked #1 in offensive efficiency last year...as our our opponent's had the best D in the NCAA (the ACC was well, kind of good defensively). Pomeroy used to have listed the raw offensive and defensive efficiency numbers, which don't take into account opponent's, but I can't find it up there on the new version of the site.

Pyth (Short for Pythagorean, like the triangle. Long story.) is the result of a formula which takes the offensive and defensive efficiency to find an expected winning percentage over a regular season.

Thanks very much, Loran16. Wanted to give you a positive comment, but it told me that I had to spread around some other comments before giving you one.

[loran16]

Thanks very much, Loran16. Wanted to give you a positive comment, but it told me that I had to spread around some other comments before giving you one.

Glad to help. (And i don't really care about the comment system, so don't worry about it.)

[MChambers]

I'm not sure I can be as dorky as Jason, but I'll try.

Duke is back to #1 in the two leading dork polls after last night's impressive win. (In Sagarin, we're still #1 in Predictor and #2 in Pure Chess, but #1 overall.)

[JasonEvans]

I found this column interesting from Ken talking about how little we know about the season and how wide a range there is for team's results at this point.

Take Boston College for example. They lost to Yale last week. Yet if the ratings are accurate right now, there’s a 3% chance BC goes 11-5 in ACC. In 10,000 runs, there was one trial where they went 0-16 and one where they went 14-2. The model isn’t saying BC is going to improve or implode in those cases. It’s the same team that is currently rated seventh-best in the ACC providing that range of results.

--Jason "to me, this column hearkens back to 1994-95, when Duke was very good early in the year and then... well... you know" Evans

[JasonEvans]

A cool function on Ken's site...

He currently gives Duke a 6.79% chance of going unbeaten in the regular season wth a 12.72% chance of going perfect in the ACC regular season.

That's about a 1-in-15 chance of unbeaten regular season, 1-in-8 in the ACC.

-Jason "insane? Perhaps, perhaps not" Evans

[ice-9]

I don't like that kenpom uses preseason bias in his rankings. I understand his premise that preseason rankings are the most accurate way to predict postseason finish, but IMO only as it applies to the top 25, or perhaps even just the top 15. Past the top 25, the preseason poll loses its predictive power and I'd prefer if kenpom just kept his rankings numerically "pure" and let his readers make adjustments upwards or downwards.

[superdave]

A cool function on Ken's site...

He currently gives Duke a 6.79% chance of going unbeaten in the regular season wth a 12.72% chance of going perfect in the ACC regular season.

That's about a 1-in-15 chance of unbeaten regular season, 1-in-8 in the ACC.

-Jason "insane? Perhaps, perhaps not" Evans

If you figure that we have 6-7 games this year where our opponent has, say, a 1/3 or better chance of beating us and we just won one of those games (K State), then our odds of going undefeated should have ticked up. If we beat Izzo's team then they should go up again a little, right?

The ACC is never as bad as the pundits enjoy imagining though.

[arnie]

Just checked RPI and Duke was 24th behind such stalwarts Appalachian State and Vermont. App has a 1-1 record. I realize its very early, but any ideas as to how this ranking system could be that far off??

[-jk]

Just checked RPI and Duke was 24th behind such stalwarts Appalachian State and Vermont. App has a 1-1 record. I realize its very early, but any ideas as to how this ranking system could be that far off??

RPI is a very different beast. It's less about what you do, and more about who you play. Or the reason VaTech missed the tourney last year.

-jk

[SuperTurkey]

RPI is a very different beast. It's less about what you do, and more about who you play. Or the reason VaTech missed the tourney last year.

-jk

Normally, I would say that our RPI will shoot towards 1 once we enter conference play. Given the state of the ACC, that may not be so true this year.

[loran16]

If you figure that we have 6-7 games this year where our opponent has, say, a 1/3 or better chance of beating us and we just won one of those games (K State), then our odds of going undefeated should have ticked up. If we beat Izzo's team then they should go up again a little, right?

The ACC is never as bad as the pundits enjoy imagining though.

KState was actually still predicted as like a 75% chance of winning for us.

http://kenpom.com/team.php?team=Duke

Most likely losses: (First Number is chance of losing, 2nd is to whom and when)
29% @UNC on 3/5
25% @FSU on 1/12
23% @MD on 2/2
20% @VT on 2/26
20% @NCSt on 1/19
14% @St.Johns on 1/30
14% @Miami on 2/13
13% MSU on 12/1 (First Home Game on list)
12% Butler on 12/4 (Neutral Game)
10% UNC on 2/9
8% Clemson on 3/2
8% @UVA on 2/16
7% Temple on 2/23
7% MD on 1/9
6% @Oregon on 11/27
6% NCSt. on 2/5
5% Boston College on 1/27
4% Miami on 1/2
4% @Wake on 1/22
2% St. Louis on 12/11
2% UAB on 1/5
2% UVA on 1/15
2% GT on 2/20
1% Bradley on 12/8
1% @UNCG on 12/29
0.3% Elon at home.

-------------------------------------
So yeah, we only have 5 games where kenpom predicts a 20% or higher chance of a loss, and non where the odds are 30%.

[JasonEvans]

RPI is a very different beast. It's less about what you do, and more about who you play. Or the reason VaTech missed the tourney last year.

The RPI is purely mathematical and is made up of your win percentage (25%), your opponents' win percentage (50%) and your opponents' opponents' win percentage (25%).

Because it is only mathematical, there is no "pre-season" bias built in. As a result, at this early point in the season when there have only been a very limited number of games, the RPI is pretty darn haphazard and in no way indicative of who the best teams are. In fact, even at the end of the year, the RPI is not designed to show who the best team is. It is designed to measure which teams have played well against a tough schedule.

--Jason "if there was no preseason bias included in the other 'Dork Polls' you would see Sag and KenPom looking out-of-whack early in the year too" Evans

[pantone287]

Early as can be, but Kansas at a very strong .9858 (above what we ended the year at last year) in Pomeroy's ratings.

The way they've played thus far plus the addition of Selby makes them a very serious source of concern. Caveats all over the place, but early results have 4 National Championship level squads in Kansas, Duke, Ohio State & Pitt and a pretty good drop off after that.

Sagarin in general agreement about nat'l contenders but is less sold on Ohio State. Sagarin also has the teams more closely bunched.

Both systems working with assumptions about strength as part of their formulas at this point.

[BattierBattalion]

Early as can be, but Kansas at a very strong .9858 (above what we ended the year at last year) in Pomeroy's ratings.

KU is definitely a contender, but they've only played cupcake teams so far. These rankings aren't going to be truly meaningful until conference games start.

[loran16]

KU is definitely a contender, but they've only played cupcake teams so far. These rankings aren't going to be truly meaningful until conference games start.

How you play against Cupcakes is a very relevant measure of success. Yes conference play can reveal true talents as well, but being able to hold weaker teams to low scores and blow them out tells you whether you're a good team or not.

So don't discount that totally. That said, there are plenty of relevant non-cupcake games on everyone's non-conference schedule. Kansas' next 6 games are against:
#12 Arizona
#46 UCLA
#21 Memphis
#90 Colorado State
#72 USC
#67 Cal.

(All rankings are Pomeroy rankings. Arizona and Memphis are at true neutral sites). KU hasn't played anyone ranked better than 120 just yet for comparison. Duke has played two (#36 Marquette and #19KSU).

If KU is still ahead of Duke after that point, it'll be very relevant. And of course, they're playing WITHOUT Josh Selby...which supposedly will be a Kyrie Irving-like impact on their team.