Advertisement

Thursday, February 19, 2015

Keegan

Column: Kansas gets RPI dance just right

Kentucky guard Andrew Harrison (5) gets to the bucket between Kansas forward Cliff Alexander (2) and forward Perry Ellis (34) during the second half of the Champions Classic on Tuesday, Nov. 18, 2014 at Bankers Life Fieldhouse in Indianapolis.

Kentucky guard Andrew Harrison (5) gets to the bucket between Kansas forward Cliff Alexander (2) and forward Perry Ellis (34) during the second half of the Champions Classic on Tuesday, Nov. 18, 2014 at Bankers Life Fieldhouse in Indianapolis.

Advertisement

Television earned the nickname “Boob Tube” because it really can turn our brains to mush. Particularly in sports, most notably in discussions comparing NCAA Tourament résumés and seedings.

Anytime a graphic pops up and compares two college basketball teams’ relative merits, it cites record vs. top-25 teams, top-50 teams, etc. Those rankings referenced are RPI rankings.

Here’s the problem: The RPI is the worst indicator of one team’s strength relative to another’s of any system, human or computer. It’s skewed way too heavily toward a team’s strength of schedule.

The formula for the Ratings Percentage Index: 25 percent a school’s winning percentage, 50 percent an opponent’s winning percentage, 25 percent the winning percentage of opponents’ opponents.

Instead of looking at a pair of teams’ RPI rankings to determine which is better, let’s compare the evidence this season for the two winningest college basketball programs of all-time, Kentucky and Kansas, in that order.

Both schools have played 26 games.

Not much upon which to go in regards to common opponents. Kansas defeated Tennessee, 82-67, on a neutral court. Kentucky won at Tennessee, 66-48. Kansas won at Texas, 75-62. At home, Kentucky defeated Texas, 63-51.

At home, Kansas is 13-0, Kentucky 16-0. On neutral courts, Kansas is 3-1, Kentucky 2-0. Just a slight edge to Kentucky there.

On the road, Kansas is 5-4, Kentucky 8-0. Big difference.

The obvious trump card: Kentucky 72, Kansas 40 on Nov. 18 in Indianapolis.

So Kentucky (26-0) defeated Kansas (21-5) by 32 points, has won all its games, and according to the RPI, KU is the best team in the country, UK the second-best. That alone should be all the evidence needed to scrap the RPI, never using it in discussions on TV, much less in rooms where NCAA Tournament selection-committee members hold meetings.

But that will never happen.

Even selection-committee members don’t know it, but the seeding process, because it uses the RPI as a tool, is as much about rewarding schools that schedule aggressively as it is about rewarding teams for what they accomplished on the court. TV, of course, and in turn TV viewers, benefit most from that reality.

You don’t like your seed? Your weak strength of schedule is to blame. So if you want a better seed next season, play a tougher schedule in November and December. That results in far better college basketball games on the tube in the first two months of the season, when the sport is competing not only with the NBA but with the NFL and college football for viewers.

Nobody better understands how to build an RPI better than does Larry Keating, KU’s Special Assistant to the Athletics Director. He’s so well connected in college basketball, he can pick up the phone, talk to a friend from all the major and mid-major conferences and get a feel for which teams figure to have good records but still aren’t likely to beat Kansas and will have a decent enough schedule to compile a good RPI of their own.

The Associated Press college basketball poll, released Monday, ranks Kansas eighth. Sagarin and kenpom.com, both of which include the most recent games in their data, rank Kansas ninth. The predictor portion of Sagarin has Kansas at No. 13.

The goal of those computer rankings is to determine the relative strength of teams, so you won’t see TV broadcasts giving much airtime to those. The goal of the RPI is to motivate teams to compile strong schedules so we don’t change the channel in the middle of an early-season rout and keep the game tuned in throughout the entire game because it appears headed for a dramatic finish. (Only “Columbo” was compelling enough to keep us glued to the screen even though we knew the outcome from the outset.)

Kansas put together the toughest schedule in America, so nobody need worry that it will be slighted come Selection Sunday.

Comments

John Randall 4 years, 9 months ago

The RPI rating is significant as a measure – the ranking very often isn't. Two or three (as many as five) teams may well have ratings so close to each other that there is no reason to consider one superior to another.

But the only time those (essentially identical) ratings are seen is in a list (ranking) which implies superiority where there actually is none, or so little as to be as useless as counting how many times a team "wins" the opening center tip off.

Conversely, two teams adjacent in that same list may well have a relatively large (therefore significant) disparity in their ratings which is hidden because the listing (ranking) shows them next to each other. Whether or not the ratings are accurate or meaningful, the presentation in a list (as a "ranking") will be deceptive a good share of the time.

Steve Zimmerman 4 years, 9 months ago

excerpt from b/r: "Even if we exclude the Champions Classic in which Kentucky went nuts and blocked 11 Kansas shots, opponents are still averaging 4.9 blocks per game against Kansas. The Jayhawks have had 13.3 percent of their two-point attempts blocked, which ranks 337th in the nation."

makes wonder how superior a team ranked high in RPI vs other top25.

Jonathan Allison 4 years, 9 months ago

TV viewers benefit because more games get televised and because the matchups are better, but TV networks benefit most by far because they get to hike up advertising prices and extend timeouts to accomodate more ads and generate more billions and billions in revenue. The TV viewer just gets to watch an extra game each week rather than watching The Bachelor or whatever other trash the TV networks are selling with all those billions they rake in from advertisers.

Ron Sterling 4 years, 9 months ago

People need to stop claiming that RPI is either a measure of which teams are best or that that's what it's claimed. RPI ranks wins versus schedule. KU has played a much harder schedule than UK and won against that schedule. Going 21 - 5 against the #1 SOS will get you the #1 RPI. UK being undefeated against their schedule at this point in the season is all the indication you need of who is the better team, but filling up on cupcakes early and playing in a much weaker conference hurts their RPI.

The reason the selection committee uses RPI is because they're trying to reward good teams for playing good schedules with good seeding. Seeding has nothing to do with which team is best. It has everything to do with which teams challenged themselves and performed well. You can argue the merits of whether or not that's how it should be, but the perception that RPI is in any way an attempt to rank the best teams or that it's inherently flawed is what needs to be scrapped.

Walter Bridges 4 years, 9 months ago

"Seeding has nothing to do with which team is best."

As my favorite tennis player often said: "You can't be serious". When the selection committee goes through the seeding process it uses these variables: S-Curve (https://bracketville.wordpress.com/s-curve/), record/polls, strength of schedule and RPI. And yes there is a difference between SOS and RPI better explained here: http://www.collegerpi.com/rpifaq.html

The entire process is to ensure the best teams are seeded first, second, third and so on. The process starts getting mucked when comparing schools not in the five major conferences to schools that are.

Using your example, it almost sounds as if Kentucky's RPI and SOS could prevent them from being a #1 seed. That won't happen. Every major poll has Kentucky the runaway best team and they will be seeded as the first #1 seed unless they completely fall apart.

Compare the seeds with the polls at the end of the season and then tell me that seeding has NOTHING to do with which team is best.

To your next comment "the perception of the RPI is in any way an attempt to rank the best teams or that it's inherently flawed is what needs to be scrapped.".

Again, I completely disagree. If that was the perception than why isn't Kansas ranked #2? I feel certain that most people understand that the RPI is helpful in determining the best team but as the polls show (again using Kentucky because you did with its comparatively low RPI) that the number one team in every poll is Kentucky regardless of its RPI. Same with Gonzaga and several other mid-majors.

Ron Sterling 4 years, 9 months ago

I thought we were avoiding each other? Has that changed? Yes, I know that you didn't reply directly to my comment, but replying to my conversation is involving yourself it in.

Ron Sterling 4 years, 9 months ago

Who said RPI and SOS are the same thing? Don't straw man. Of course they aren't the same, but RPI derives, in part, from SOS. The better SOS you've played and the better you do against that SOS, the higher you'll rank in RPI. That's why KU is over UK in RPI even though we all know UK is the superior team, both in record and head to head. Also, UK doesn't have a comparatively low RPI. They're #2. But the reason they aren't #1 is because they played a lot of cupcake games and play in a much weaker conference. The reason they aren't lower is because, despite significantly lower SOS (which is still top 30), they played enough marquee match ups and have gone undefeated.

RPI does NOT attempt to rank the best teams. As you pointed out, KU ISN'T #2 in the AP/Coaches polls because those polls DO attempt to determine the best teams. And fine, perhaps I'm being a little hyperbolic in stating that the committee doesn't care about what teams are the best, they do to a point, but they reward good teams that performed well against good schedules more because it helps their bottom line.

SDSU had a great record a couple tournaments ago and still got a 2 seed because despite being high in the polls when the tournament was seeded, they didn't do as well as some other teams based on RPI and SOS.

Regardless, almost every point you bring up reinforces my original thesis: RPI isn't an attempt to rank the best teams, it's an attempt to ranks which teams racked up the most wins against other good teams, and the selection committee relies on it heavily because they want to encourage that kind of scheduling.

Shane Johnston 4 years, 9 months ago

Yes, the RPI clearly leans toward strength of schedule. How well people understand it doesn't matter too much - the important thing is that the selection committee relies on it. College basketball is better when every team faces a strong incentive to schedule good teams. That incentive is effective as long as we rely on the RPI for seeding. College basketball is less popular today than auto racing and hockey. Maintaining the incentive to schedule well is one way to help pick the sport back up.

Walter Bridges 4 years, 9 months ago

I hope you are wrong about hockey and auto racing Shane. Does that cover the entire season or just the NCAA (which I've always believed to be a top draw)?

Shane Johnston 4 years, 9 months ago

The entire season. Nationwide Harris poll

Sean Swindler 4 years, 9 months ago

I am hoping this all comes back to bite KY - they will obviously have the overall #1 seed - but after playing in the super-weak SEC, hopefully they can play their 2nd round against, oh, West Virginia (which is and 8/9 seed) and it will likely be the toughest team they have played since December.

Really love to see Baylor as a 4 or 5 seed get KY in the Sweet 16

Tom Jones 4 years, 9 months ago

West Virginia is on probation, but I'm with you in hoping they run up against a bad matchup early.

Bryce Landon 4 years, 9 months ago

As far as I know, WVU's probation does not include a postseason ban.

Walter Bridges 4 years, 9 months ago

You've got it right Bryce.. If it did include a ban on postseason there would be no need to worry as to their seeding.

Cameron Cederlind 4 years, 9 months ago

The issue is when you rank Northern Iowa a 9 seed in the tournament when kenpom has them ranked #19 and should have been a 4-5 seed.

John Pritchett 4 years, 9 months ago

That's a good point. If you accept that RPI is not a ranking of the team's relative abilities, then any attempt to use it to reward a team for a strong schedule is merely an illusion. Because if you punish a really good team for a soft schedule, you're actually under-seeding them and you end up putting that team against a team you're supposedly rewarding. What good is it to be a 2 seed when you end up playing a 15 seed that could have easily been an 8? In other words, it's fubar.

Anyway, it's been obvious for a long time that seeding in the NCAA tournament is a crap-shoot. And success in the tournament has as much to do with luck as skill. But it's still the greatest sporting event ever.

Rodney Crain 4 years, 9 months ago

All Hail Larry Keating, KU’s Special Assistant to the Athletics Director.

Nothing wrong with playing the game well., the game of RPI and how it works with the selection committee.

I agree we are a 13th ranked team right now, we will get a 2 or 3 seed. Nice article Tom K.

Rick McGowwan 4 years, 9 months ago

Keegs takes a lot of negative comments on this board, some right and some wrong, but the last week we have to admit that he has been on FIRE.

Great stuff.

Commenting has been disabled for this item.