Tuesday, March 11, 2008

Completely Irrelevant RPI Data (To the Pollsters) Final 08 Edition

So, who does better job of determining the strength of teams? Sports writers? The secretaries for college basketball coaches? The RPI? Completely Irrelevant RPI Data is here to help you decide. And as always, do not miss Goro’s Rants, especially this gem of a comment this week:

The definition of a good coach, in an ideal world, should be first and foremost, ahead of any win totals, the ability to graduate his players. Something like 45 out of 48 players under Hinson's watch will have graduated and are now a part of the Missouri State fraternity.

Unfortunately, in today's win-at-all-costs mentality, a coach only gets a finite time to get his team to the NCAA tourney before the university heads get in a snit and pull the plug. It didn't matter that Barry's AVERAGED 19 wins per year, or that he's become such a well-respected dignitary for the conference.

I put all the blame on the insanity and moronity of the NCAA selection committee nitwits during Hinson's tenure in not inviting three very deserving Missouri State teams to the NCAA tourney (2000, 2006, and 2007). Their obvious BCS vanity is by far the biggest reason why Barry's not coming back, and that's really sad. My admonition for the cognitively-challenged MSU bigwigs is this: the new court's got some nice amenities, but you'll be lucky to keep any new coach around another nine years like you just had with Barry Hinson.


The Hall of Shame


[Team Record Conference (AP ranking, Espn/USA ranking, RPI ranking) Difference between poll average and RPI]

#1 Purdue 24-7 BE (17,16, 37) +20.5

Two weeks ago, Purdue was over-ranked at #16 and #19. Over the past two weeks Purdue beat mediocre Minnesota and awful Northwestern at home. They followed that up by losing on the road to an Ohio State team which is on or around the bubble, before somehow managing to struggle on the road at Michigan before ultimately pulling away. The pollsters managed to reward this series of mediocre performances with an average one spot rise in the polls. To play the broken record, the Big 10 is just not that good this year. Beyond a couple wins against Texas, Big 10 teams had a distinctly undistinguished out of conference record, and Purdue is a classic example. Not only was Purdue defeated by #169 Iowa State on a neutral court, but they also lost AT HOME to #194 Wofford.

#2 Davidson 23-6 Southern (23,25, 41) +17

Davidson falls out of the top spot, but maintains its spot in the Hall of Shame after a week in which it continued to beat up on awful opposition and was rewarded with an average three spot rise in the polls. Lets play blind resumes. Team A has a 26-5 record, which includes a 3-2 record versus the RPI top 50 and a 4-3 record versus the RPI top 100. Additionally, one of Team A’s losses came on the road in double overtime to a ranked team. Team B has a 25-6 record which includes a 0-3 record versus the RPI top 50 and an 0-5 record versus the RPI top 100. Team B played two ranked teams close at home but lost in regular time in both instances. Which team deserves votes in the polls? Which team deserves an at-large bid to the NCAA tournament? Team A naturally, which happens to be South Alabama, which received a total of 53 votes in the two polls this week. Team B is Davidson, which received a total of 241 votes in the two polls this week. Both teams need wins in their conference tourneys to be sure of spots in the dance, but should Davidson lose, they can count on a trip to the NIT, unlike South Alabama, who would still be in with a shout.

Davidson supporters – if you comment here, kindly put those wonderful educations to good use and produce some solid analysis on why I am wrong.

Editorial Note: This was written before last night’s games, which resulted in Davidson winning the Southern Conference tourney, and South Alabama losing in the semifinals of the Sun Belt tourney. Best wishes to Davidson in the quest to validate your ranking with a Sweet Sixteen run. This would, naturally, prove me wrong. Best of luck to South Alabama in your quest for an at-large bid. You represent one of the more interesting bubble cases this year.

#3 Gonzaga 25-6 WCC (20,22, 30) +9

Gonzaga climbs back into third place in this weeks Hall of Shame after two wins against an awful Santa Clara team resulted in a rise in its ranking in both polls. QUE? Gonzaga has one signficiant victory since December 1, on March 1 at home against St. Mary’s. Throughout December, January, and February, Gonzaga developed a shiny record based on wins against patsies while losing to the quality teams it faced. This team is NCAA-worthy but not ranking-worthy, and at some point the afterglow from the Elite 8 run NINE years ago needs to wear off. . .

#4 Butler 28-3 HORZ (12,10, 18) +7

After a couple weeks on the sidelines, Butler re-joins the Hall of Shame after it rose an average of two spots in the poll on the strength of a neutral-court victory over, drum roll please, ILLINOIS CHICAGO! WOW! HOW STUNNING! As in past weeks, the best illustration of Butler’s over-ranking is a direct comparison with Drake. Drake is a team which beat Butler on its home court on February 23rd, but continues to languish well behind (#16, #18 this week) in the rankings. Butler’s best wins came against #49 Ohio State, #57 Virginia Tech, #59 Texas Tech, #62 Southern Illinois, and #64 Florida State, with all of these coming in November and December. Drake’s Best wins came against #18 Butler, #31 Illinois State (X3), #51 Creighton (X2), and #62 Southern Illinois, with the latest of these coming on Sunday. The only category in which Butler can claim the advantage is the arena of bad losses, of which it has none, unlike Drake, which has two (#102 Bradley and #124 Missouri State). All told, Butler is evens at best with Drake. It will be interesting to see how far these two teams progress, if at all, in the tournament.

#5 Stanford 24-6 P10 (11,11, 17) +6

You heard it hear first: “Stanford is not top ten material - at least four teams, Xavier, Wisconsin, Georgetown, and Texas deserve to be ranked higher. This will be established when Stanford loses at UCLA on Thursday, and struggles on the road at USC on Saturday.” Let us remind ourselves of Stanford’s out of conference strength of schedule (#296) which was achieved by playing the following patsies at home: #183 Yale; #188 Santa Clara; #223 Northwestern State; #237 Fresno State; #273 Colorado State; #308 Harvard. They rounded this off with #335 Sacramento State, a 4-24 team whose best win came by one point, at home, against #231 Montana State.

Dropped from rankings:
Mississippi State (now unranked)


The House of Pain

[Team Record Conference (AP ranking, Espn/USA ranking, RPI ranking) Difference between poll average and RPI]

#1 Oklahoma 21-10 B12 (33,NR, 25) –9.5

I did a double take when this one popped up. Oklahoma was left for dead at 3-5 in the conference, and 15-8 overall after an awful loss AT Colorado State on February 9. Since then, they have won six of their last eight, including wins against Baylor, Texas A&M, and Texas Tech, and have made themselves a team worthy of an at-large berth without most people noticing. This is not to say, however, that they are deserving of top 25 spot. . .

#2 Vanderbilt 25-6 SEC (18,17, 10) –7.5

Vanderbilt remains the most consistently disrespected team from the big conferences, although it is difficult to find any fault for their drop in the polls this week given that they lost at ALABAMA to close out their regular season. . .

#3-T Indiana 25-6 B10 (22,20, 16) –5

*GIGGLE* Indiana becomes the first team in the history of this feature to appear in both the Hall of Shame and the House of Pain in the same season. But frankly, this team is clearly not playing to the same level under Dakich as it had under Sampson, and the loss at Penn State on Sunday is a brilliant illustration. At the rate things are going downhill, Indiana may be happy with a second-round exit, as opposed to the Final 4 that was being predicted a few short weeks ago.

#3-T Drake 28-4 MVC (16,18, 12) –5

See above under Butler.

#3-T South Alabama 26-5 Sun Belt (28,30, 24) –5

See above under Davidson.

Dropped from Rankings: Arizona (out of RPI top 25); Pittsburgh (out of RPI top 25)


Average Conference (or State or Category) Bias:

Media Darlings: +11 (Butler, Gonzaga, Davidson)
Teams from Indiana: +6.9 (Butler, Notre Dame Purdue, Indiana,)
Big 10: +4.4 (Wisconsin, Purdue, Michigan State, Indiana)
Pac 10: +2.8 (UCLA, Stanford, Washington State)
All Others: +2.4 (Memphis, Xavier, Butler, Drake, Gonzaga, Brigham Young, Davidson, South Alabama)
Big East: +.1 (Georgetown, Louisville, Notre Dame, Connecticut, Marquette)
ACC: -2 (North Carolina, Duke, Clemson)
Big 12: -2.8 (Kansas, Texas, Oklahoma)
SEC: -5.25 (Tennessee, Vanderbilt)

After several weeks of constricting biases, things went in the opposite direction this time, for the most part because there are certain teams, like Purdue, which are rarely penalized for losses as they should be, and others, like Vanderbilt, who do not get the benefit of the doubt. The action around Purdue meant that the Hoosier love-fest continued. On other fronts, the ‘Media Darlings’ continued to attract suitors. . .


Teams with closest correlation between ranking and RPI:

1. Memphis +.5
2-T. North Carolina, Connecticut, Washington State (+1, +1, -1)
5. Marquette (-1.5)


Teams included in analysis:

[Any team in the top 25 of the AP Poll, the Coaches Poll, or the RPI.]

North Carolina 29-2 ACC; Memphis 30-1 CUSA; UCLA 28-3 P10; Tennessee 28-3 SEC; Kansas 28-3 B12; Texas 26-5 B12; Wisconsin 26-4 B10; Duke 26-4 ACC; Georgetown 25-4 BE; Xavier 26-5 A10; Butler 28-3 HORZ; Stanford 24-6 P10; Louisville 24-7 BE; Notre Dame 24-6 BE; Connecticut 24-7 BE; Purdue 24-7 BE; Drake 28-4 MVC; Vanderbilt 25-6 SEC; Michigan State 24-7 B10; Gonzaga 25-6 WCC; Indiana 25-6 B10; Washington State P10; Brigham Young 25-6 MWC; Davidson 25-6 Southern; Marquette 22-8 BE; Clemson 22-8 ACC; South Alabama 26-5 Sun Belt; Oklahoma 21-10 B12.

6 comments:

Anonymous said...

RPI is a ridiculous measure. Sagarin has Davidson ranked 26th and is typical of most computer rankings. The statistical superiority of Sagarin over the RPI is well-established. Your "hall of shame" evaluation is totally a function of using a woefully inadequate metric for comparison with the polls.

To demonstrate how silly the RPI is, compare the NCAA's ranking for Davidson (presently 41) with the one at Realtimerpi.com (presently 31). The difference? Davidson's game at UCLA is considered neutral by the NCAA and an away game by realtime. The games with UNC and Duke at Bobcats Arena in Charlotte are considered home games by the NCAA and neutral by realtime. Realtime recognizes reality since Davidson fans were a distinct minority at both games.

Any metric which fails to recognize the reality that playing UCLA in LA is an away game for the Wildcats has no place being used by anyone interested in accuracy or fairness.

Point your finger of shame in the mirror.

Profane said...

Anonymous,

Emotion often leads one to engage in juvenile personal attacks. Leave them at the door if you wish to comment here in the future. Shame on YOU.

Emotion is also the enemy of any good argument. In particular, yours fails to take account of the fact that one of the functions of this series is to cast a critical eye on the RPI. Arizona and Dayton, for example, remained in the top 25 of the RPI long after any rational basketball person would have placed them in a top 25 poll. The pollsters were also correct in ranking Notre Dame at a time when they were still in RPI purgatory.

As for the Sagarin rankings, their strength, taking into account margin of victory, is also their weakness. A win is a win, whether it be by one point on a buzzer-beating tre or by a thirty-three point blowout. Close counts in horseshoes and hand grenades, but not in basketball. Furthermore, Sagarin, like most computer rankings, rewards teams for running up the score long after a contest is finished. The "strength is weakness" argument works in the opposite direction for the RPI, which takes into account the result, the opponent, and the location (on which their does need to be some fine tuning on so-called neutral sites), but NOT the margin of victory.

Sagarin ratings are a predictive measure. The RPI is a rough measure of achievement. It is an apples to orange comparison. Asserting that "The statistical superiority of Sagarin over the RPI is well-established" is meaningless unless the goal of the ratings is addressed.

Anonymous said...

So you defend the ridiculous fiction? Because without embracing the fiction of home/neutral used by the NCAA, even the RPI would have Davidson at 31 and your differential with the polls would diminish to irrelevance.

Profane said...

Anonymous,

Straw man argument. Try again.

Anonymous said...

Go back and re-learn your logical fallacies. That's not a straw man argument. You'll have more credibility, if you respond to the point made.

RPI is a badly flawed measure in general. Its specific application this year with respect to Davidson is exacerbated in its error by adjustments made by the NCAA.

Your hall of shame is predicated on the assumption that the RPI is an accurate measure of the quality of the teams. That's a ridiculous assumption, but even within the parameters of your assumption, Davidson's poll rankings don't vary much from its RPI -- unless that RPI is contaminated by adjustments which defy reality.

In sum, your analysis rests on a critical assumption. Evidence and argument which assail the credibility of your essential assumption do not constitute a straw man. That's ridiculous for you to assert.

Profane said...

Anonymous,

Your argument relies on mis-statement of my position. That is a textbook straw man argument.

It was inherent in your first post, yet I gave you a pass in the hopes that you might, indeed, be interested in a serious argument.

Given that your latest comment merely represents a re-statement of your past posts coupled with yet more incivility, including personal attacks, I am not going to bother with a further response on the merits. Best wishes.