Blog Entry

RPI still hovers over, cloaks selection process

Posted on: February 18, 2012 1:30 pm
Edited on: February 18, 2012 1:52 pm
The room where it all went down Thursday and Friday. We were secluded in secrecy, soft drinks and statistics. (NCAA)

By Matt Norlander

No matter how emphatically and repeatedly the NCAA Selection Committee insists the RPI isn’t a major factor in picking and seeding 68 teams into the greatest sporting event the universe has ever known, that’s simply not the case. The RPI, as flawed a tool as any mainstream collective ranking metric we have in college basketball, still stenches up the process like burnt popcorn.

(If you’re still not sure why the RPI is built like a Popsicle-stick castle, I’ll promptly point you in the direction of this and this and this. Get to learnin'.)

I had the luxury and pleasure of attending the NCAA mock selection meetings in Indianapolis Thursday and Friday, at the Conrad Hilton. My hope was to file a couple of quick blog updates/entries during the process. That’s nearly impossible. We started at 1 p.m. on Thursday, worked until about 11, then were at it again at 8 a.m. Friday morning and went until 2:30, and then it was off to the airport. It was a beautiful, dream-come-true of a grind. So here I am, frothing to share with you the events of the past 48 hours. On Sunday, I'll have a piece on seeding and bracketing, my loves and laments.

By the way, here's the field of 68 we concluded on, as if the season ended Wednesday night. I'll have more on this in tomorrow's piece. The irrational responses -- some of them tongue-in-cheek -- on Twitter made the entire process worthwhile. If only Jay Bilas had backlashed at us, then it truly would have been a Selection Sunday simulation.

Off camera, NCAA director of media relations David Worlock spellbinds us with his humor and deflection of the issues. (NCAA)

Now, let’s return to the RPI. Here’s my thing. I don’t have an issue with it being in the tool belt. It’s probably always going to be something that factors in, and I’m just going to have to live with that like I live with the cowlicks that command my hairstyle. The RPI generally organizes teams in a reasonable way when it comes to clustered arrangement of the best-to-worst tiers of teams. That said, why can’t the RPI get treated like every other rankings system: equally? Right now, it’s not. Right now, as it's been for the past 30 years, despite semi-annual adjustments to the bare-bones formula (that's what makes it bad and manipulable),  the Ratings Percentage Index remains the favorite flavor to savor for picking the "best" teams into the field.

The NCAA allows (but from what I interpreted, does not heartily endorse) any Selection Committee member to use Sagarin, KenPom, LRMC, Massey or any type of ratings system (including — WHAT — the Coaches’ Poll? It’s true, unfortunately). Those systems are not brought up on the big screen, unless by request, which never happened at our mock.

I'm guessing its seldom a Pomeroy team page will get clicked to the projector screen this year, too, especially since it's now subscription-based, and that $20 annual fee might be a touch too much.  

The RPI is the peanut butter that keeps the primary data smudged together for the Selection Committee. It still permeates the process. And the NCAA still wants to deny that. The NCAA likes to say bringing up a team’s specific RPI ranking doesn’t often come up when debating two teams’ inclusion or seeding. While that’s true, from the outset, the organization, presentation and general data on a team is dressed up in an RPI shirt with an RPI hat and a cute pair of RPI gloves.

On a few occasions, NCAA tournament Poobah Greg Shaheen (who, along with 2012 Selection Committee chair Jeff Hathaway, was awesome) would say, “OK, how many times did you find yourself talking about RPI in the past 10 minutes?” or something to that effect. No one responded with:




"A halfway decent amount!"

But that's not the point, because, as if you're being hypnotized into a train of reason and deduction, the RPI is placed right in front of the committee members’ faces from the start of the process, and I sincerely doubt they deviate from the materials and data given to them by the NCAA and its computer sorting/ranking/bracketing/filterin
g system (which is a slick, impressive computer program). This year, the NCAA has made public for the first time its Nitty Gritty (yes, that’s a capital N and G) sheets. These sheets rank teams by RPI. Immediately, you’re sorting teams in accordance with a flawed system.  Within the Nitty Gritty you’ll see nine of the 16 columned categories are RPI-dictated.

It doesn’t stop there. On team sheets and in side-by-side comparisons, the only metric numbers available are RPI. It's very easy to use the data baked into the NCAA's team sheets and use that in addition to eye test discussion to draw conclusions. In such a scenario, which is one that occurred over and over and over at the mock, you're being unfair and myopic to the process. And more than inclusion to the field, you're jeopardizing fair and realistic seeding -- something, again, I'll get to in Sunday's post.

Why not rank teams with a median of four, five or six respected, mainstream rankings systems, such as the ones listed above? The reason the NCAA doesn't is because some of the systems account for future results, and they don't want any predictive measures entering into the process. I say: the RPI is only one net, and so you've got many holes. The more nets you throw on top of it, the more reasonable general conclusion you can come to, and so fewer and fewer holes are possible in the rankings system. Keep the Nitty Gritty, but tweak:

  • An average ranking tally from RPI, Sagarain, LRMC, KenPom, Massey, and maybe even the brand-new BPI. This will be your master ranking.
  • A neutral-court record. It's glaringly absent from the Nitty Gritty, yet it's part what the NCAA tournament is all about: winning games on neutral courts. 
  • Get rid of conference record. Those are on team sheets and are not paramount to the grand overview the Nitty Gritty aspires to be.
  • Instead of "Record against RPI 1-25, 1-50" etc., give a record against the conglomerate. More inclusion from all systems eliminates the RPI's influence over the Nitty Gritty and general impression committee members glean when absorbing all this information.

I’ve been talking in geek speak for a few grafs here, so let me stop for a second and emphasize that I are pretty much everyone else relied heavily on instinct and eye tests, too. That’s a large part of the discussion. Recalling when a team won or lost and what those circumstances came down to. Debating a team’s merit based on its body of work, its best wins and worst losses: all of those things were a backdrop to the shirts vs. skins question.

Beyond all else, right now, who do you take on a neutral court in a shirts against skins game. Go with your hunch if data is too overwhelming and inconclusive.

Another scenario/question the NCAA said gets brought up frequently during the real selection process: Which team would you rather face in the tournament if it started tomorrow? Whatever team looks more appealing, the other one should be the pick. I liked this. Pragmatism helps, and it's good to glance away from the computer screens and go eye to eye with others when breaking down the bracket.

During the process, my partner, Rush the Court founder and EIC Randy McClure (that's him in the indie-rock glasses in the photo above) found ourselves constantly discussing a team’s merit, but also referencing Pom, Sag, LRMC and the NCAA-organized team sheets. We compared rankings and looked for outliers in the process. It was meticulous. We were frequently the last ones to submit our votes for teams as we whittled down the field (this became fodder for Shaheen to tease us), and the reason was we wanted to be thorough.

From that, here’s my conclusion: I don’t think NCAA Selection Committee members are using all available, valuable tools when picking and seeding teams. I think it’s too much information; there’s too much discussion, and the ease of the NCAA team sheets and the debates that come with it are too easy to cling to/subconsciously rely on. You have to open separate web browsers and constantly click band and forth between rankings systems. Why do that when you’ve got basic — and flawed — data in front of you that’s brought up simply by asking, “Can we get a side-by-side of Cincinnati and Middle Tennessee State?”

The process is tedious, because it needs to be, but it's still not completely all-inclusive. We'll know the NCAA truly respects other rankings systems and isn't subjective to the RPI before all else when it takes the initiative to include other credible, established rankings systems into its team sheets, computer program and debates. 

I happily admit, though, for me, whenever it's too close to call, I'll always go back to shirts vs. skins. Who you got?


Since: Jan 22, 2007
Posted on: February 18, 2012 10:17 pm

RPI still hovers over, cloaks selection process

Just to follow up on that: in all reality, once you get into the seeds below #8 (and realistically, the line's higher than that), you're getting teams that aren't going to win it all anyway.  Might as well be equitable, avoid the sins of the B(C)S, and give every team in the division a clear path (however unlikely) to a title, rather than exclude teams in favor the dregs of the bigger conferences who don't have a practical chance of winning, either.

Since: Jan 22, 2007
Posted on: February 18, 2012 10:11 pm

RPI still hovers over, cloaks selection process

A #15 or #16 seed has NEVER won an NCAA game. Never.
#15 seeds have won four games in NCAA tournaments.

The beauty of the basketball tournament is that EVERY team in Division I can have a shot at making it to the big time; win your conference, get into the dance, and then it's all on you.  I'd rather see the Ohio Valley winner have his day than put in a 10th Big 10 team to face the 11th ACC team.  Auto-bids should stay, period.

Since: Dec 1, 2006
Posted on: February 18, 2012 9:02 pm

RPI still hovers over, cloaks selection process

Here we go again...    Pomeroy and Sagarin have stated they're "predictive systems".. they are not RANKING systems.

The BPI?   HELL NO..   You rank a team higher that loses, because, oh, they've got a player who was suspended, hurt, ejected... but "otherwise" they would have won.    You did realize the BPI says "we include score differentials.. unless we dont' want to"""". &n
bsp; That's what makes for a good subjective choice?   

Good flippin' grief...  You've got ranking systems that are based on previous successes... And it don't include some of those ranking systems you guys keep popping out there.

The RPI was "good enough" in the past, all of a sudden you guys have problems iwth it being ONE of the tools... Hell you want to see something scary? 

Here is one of the items that the NCAA states is included in it's selection criteria: 
Among the resources available to the committee are complete box scores, game summaries and notes, pertinent information submitted on a team’s behalf by its conference, various computer rankings, injury reports, head-to-head results, chronological results, Division I results, non-conference results, home, away and neutral results, rankings, polls and the NABC regional advisory committee rankings

I can see it now... "Yeah, UConn is only .500 against the Big East, but hey, we're the important Big East!  So you have to include us" is in the information.. .. and some selection person says, "Hey, we have to include this letter from the Big East... So UConn is in!"

(not to pick on the Big East.. i'm just picking a well known-usually good team/conference.)

Or the "Various" computer rankings... that means, hey, i dont like the guy who made this computer program ranking for some personal reason, so were not including them for any consideration...

Or my thrid favorite... POLLS!!  In other words, you're wanting to base your selection on teams that other coaches/reporters may have watched one time ever.. or maybe someone they HAVEN'T EVEN SEEN!

Stick with the numbers that you can quantify without these ridiculous 'exceptions"...

Since: Nov 29, 2006
Posted on: February 18, 2012 7:17 pm

RPI still hovers over, cloaks selection process

Have said it over and over. There is something wrong with a tournament whereby the NIT is better than half the NCAA field every single year. If there is going to be 64 teams, give us the top 64. The tournament was so much better when it only included the Top 32. There were no puds in the field. Talk about upsets. Every team was capable of beating anybody in the field. There were no throwaway games.  The 32 team field made it impossible for the NCAA Selection Committee to "hide" top seeds from teams that were bad match ups for them (see Duke). The "usual suspects" didn't have their skids greased to the Final Four. The expansion to 64 teams wasn't done to throw a bone to 32 teams that weren't anywhere close to the Top 50. The dirty little secret, is the money. The 64 team fields makes it much easier for the committees and TV networks to shape the Final Four to get their desired result. The networks love to see a #12 beat a #5, because it clears path for one of the top seeds. They like to see Cinderella take down a #5, but they NEVER want to see one of the top seeds taken down by Cinderella. What happens now is that all of the teams ranked #9 through #25 take each other out. The seeding process is overtly political and designed to acheive a desired result. Example: Duke will seldom have teams in their region that have any chance at beating them. The higher seeds in the Duke bracket will generally be softer style teams. You may say "all the brackets have an equal amount of seeds". Do they? "Soft" teams like Gonzaga are always inflated, and given a ranking that is way too high. When the seeding is done, that "soft" high seed team is inserted into the Duke bracket, while teams like Wisconsin, Michigan State, Kansas State, Louisville, or West Virginia are seeded in brackets with teams that match their style of play. Don't believe it? Wait and look at the brackets. It is a stone cold lock that Gonzaga will be in the Duke bracket, probably with a Vanderbilt or west coast daffodil. Jerry Palm knows it, and guess where he has Gonzaga. The power, bump and grind teams get lumped together, whereby only one or two can make it out. Murray State? Look for them to be overseeded, and used to "fill" a bracket for a soft #1 seed. In the current format, half of the field isn't in the Top 50. There should be no automatic qualification for Podunk conferences. A #15 or #16 seed has NEVER won an NCAA game. Never. Never will. But yet we send good teams that deserve to be in the NCAA field to the NIT every year, so Idaho Polytechnic players can feel better about themselves. It it's got to be 64 teams, then make certain it's the 64 best teams.

Since: Jun 5, 2011
Posted on: February 18, 2012 7:10 pm

RPI still hovers over, cloaks selection process

First of all, a seed of four or better has won the tournament 80 percent of the time in the last twenty years.  In other words, nobody who actually has a chance to win the tournament is ever left out unless they are ineligible for NCAA infractions.  If there were only 16 teams in the tournament, then people would have a reason to complain, but when 68 teams make it in, those on the fringes really don't have a chance to be much more than a first round tomato can anyway.  

I like the RPI because it treats teams as numbers and doesn't have to worry about insulting a crony.  There is no homerism in the RPI, and no bias based on exposure or personalities.  If anyone is really upset, though, why not just let everyone in and play the first two and a half rounds on campuses during "conference tournament week?" That way, nobody can ever whine about not getting in.  Also, the NIT would no longer be needed, as if it is anyway.  In other words, they could kill two birds with one stone.

Since: Dec 12, 2007
Posted on: February 18, 2012 4:27 pm

RPI still hovers over, cloaks selection process

How was Valpo not in the tournament when they are the #1 seed in the Horizon League giving them Homecourt Advantage in their conference tourney when Milwaukee is in the tourney and they arent even in the top 4 of the conference? Makes no sense.

Since: Aug 21, 2006
Posted on: February 18, 2012 4:05 pm

RPI still hovers over, cloaks selection process

Kindly point out the last team that was robbed of a sure championship because of flawed seedings based on the RPI.  At the end of the day, you have to beat all the teams in your path, plain and simple.  RPI doesn't mean anything out on the floor.  I don't really give a crap about the last four teams let into the tournament, they aren't going to win it all anyway.

Since: Jan 3, 2009
Posted on: February 18, 2012 4:03 pm

RPI still hovers over, cloaks selection process

I really like the idea of including Pomeroy, Sagarin, and other rankings as well as the RPI; in this case more approaches to the data are better.  I think AWAY games are much more important than neutral sites, since they give a good idea whether a team can truly compete (I think Palm penalized Iowa St for no big AWAY wins in his Bracket, and the big AWAY win by Kansas St, who he seeded instead, over Baylor today makes him look like a genius!)  
I really do not like the idea of pooling 1-25 and 1-50 records; I would even further dissect the Top 25 into 1-10 and 1-25.  Although there is not a lot of data, if a team has gone 1-1 against Top Ten teams, that is much more impressive to me than going 1-4 against Top 50 teams (and 0-2 against the best).  Teams should be rewarded for beating Top Ten teams since these are better than most of the teams in the Tournament.  I have no problem looking at a conglomerate record of 25-100, since those teams are much more interchangeable than the 1-50 teams , which I can see fallling into 1-10, 11-25, and (25-99) bins.  I also do not like penalizing a team for their wins against teams with RPIs > 100  or 200 (Murray St and Harvard come to mind).  If they LOSE against such a weak team (and most teams do not go through their conference schedule undefeated due to rivalries and familiarity), they should be penalized, but a good tourney bound team should beat most teams ranked 25 -200.

Since: Dec 23, 2007
Posted on: February 18, 2012 3:32 pm

RPI still hovers over, cloaks selection process

the bloggers premise is wrong at the start.. the analysis is then flawed and false... what was the point of printing the opinionated garbage as fact...

Since: Dec 21, 2009
Posted on: February 18, 2012 3:24 pm

RPI still hovers over, cloaks selection process

The process will always be flawed because of the human element in the selection. Questions of bias, etc. It's almost best to let a computer select the At-Large participants to avoid accusations and controversies. By the way, Southern Miss is better than a #10 seed. See what I mean. Everyone on this Committee is probably a college graduate, loves basketball, and would like to see their alma mater advance. 

The views expressed in this blog are solely those of the author and do not reflect the views of CBS Sports or