Why Kansas is No. 1 in RPI, and why that doesn't (and RPI shouldn't) matter

RPI -- Ratings Percentage Index -- is one of the most important artifacts in college basketball. It helps to determine a team's NCAA Tournament chances, and seeding, and generally simplifies the Selection Committee's work -- rightly or wrongly.

But what RPI is not is a power rating. And yet, year after year, we get reactions like the one from Kentucky beat writers Monday, in response to three-loss Kansas being ranked No. 1 in RPI, ahead of the Wildcats (and undefeated Virginia).

Kansas ahead of UK in latest RPI deserves a big so-what? Torture numbers enough, you can get them to say anything. Meaningless and inane.

— Jerry Tipton (@JerryTipton) January 26, 2015

If on any date a computer puts a team with a 25-pointer against Temple among its three losses ahead of two unbeatens, throw it out. The end.

— Kyle Tucker (@KyleTucker_CJ) January 26, 2015

It's actually kind of impressive both Tipton and Tucker managed to find vitriol for the RPI that has little to do with its actual flaws. Tipton's "meaningless" criticism couldn't be more wrong: RPI is and has been essentially the Selection Committee's go-to metric for determining a bubble team's worthiness of being given an at-large bid for years, and is often the deciding factor in seeding. And Tucker's "throw it out" in regards to a "computer" sounds like the old anti-Bowl Championship Series rhetoric, which was usually founded in misunderstanding of what the formulas factored into the "computer ratings" of the BCS really were.

Let's tackle that latter part first, and try to clarify. RPI is not a "computer": It is a number, derived from a formula, that rates a team. (Sure, it's complex enough to require a computer to calculate it, but it's not "a computer.") That formula -- at least in college basketball-- is a team's winning percentage times 0.25, plus its opponents' winning percentage times 0.50, plus its opponents' opponents' winning percentage times 0.25.

Given the difficulty of winning on the road in college basketball, the NCAA uses an adjustment to account for that in winning percentage: A home win only counts for 0.6 wins, while a road win counts for 1.4 wins. (Neutral-site games count for a full win or loss.) And those wins are tallied and then divided by the number of wins available to determine winning percentage.

Kentucky, obviously, has a better "winning percentage" than Kansas does: Kentucky's is 1.000 (15.4 "wins" divided by 15.4 possible wins), and Kansas's is, by my calculation, 0.776 (13.2 "wins," counting a game against Utah in Kansas City as a home win, divided by 17.0 available). And that margin is actually larger than the difference between the two teams in raw winning percentage -- Kentucky's 19-0 mark obviously keeps the 'Cats at 1.000, while 16-3 Kansas would be at 0.842.

But that adjusted winning percentage is only half as important to RPI as opponents' winning percentage, and exactly as important as opponents' opponents' winning percentage. This is where Kansas shines: Its "strength of schedule" (SOS) -- which includes a game against Kentucky that the Jayhawks lost by 32 points -- is No. 1 nationally.

And it's not even close.

Kansas's strength of schedule -- which is just the last two elements of the RPI, and which you'll also see described as "two thirds a team's opponents' winning percentage (with games against said team stripped out), one third opponents' opponents' winning percentage" -- is an incredible 0.6586, more than four hundreths greater than No. 2 VCU, which boasts a stellar 0.6174 mark. The gap is larger between Kansas and VCU than the one between VCU and Villanova, which ranks No. 32 in SOS with an 0.5774 mark.

That strength of schedule could be historic, not just astronomical. Last year's Kansas team played what was eventually seen as one of the toughest schedules in the history of the sport (it produced a final strength of schedule of 0.6244, monstrous at the time), with non-conference games against Duke, Florida, San Diego State, and Villanova, and a murderous Big 12 gauntlet to run. This year's team, which played Florida, Georgetown, Michigan State, and Utah in non-conference play in addition to its game with Kentucky, is already on pace to face a much harder slate -- and the Big 12 is even better this season.

That is what's buoying the Jayhawks, who hold a very slight edge over Kentucky -- somewhere between three and five thousandths of a point -- in RPI at the moment. (Another confusing thing about the RPI: The NCAA keeps an official RPI standings, but does not release its raw numbers -- so sites like ESPN, CBS, WarrenNolan.com, and RealTimeRPI all calculate RPI on their own, and often come up with slightly different numbers, for whatever reason. All of those reports, including the NCAA's, have Kansas leading Kentucky for now.)

So, yes, Kansas is ahead of Kentucky in RPI, and it makes sense: Kansas has played an absolutely ridiculous schedule, and is being richly rewarded for that. But no one, not even the Selection Committee that consults RPI like gospel, thinks that Kansas is better than Kentucky.

The 'Cats' 32-point destruction of the Jayhawks in November was pretty convincing, yes, but Kansas has also lost to Temple by 25, as Tucker mentions, and has struggled at times this year, especially with teams like Florida and Michigan State that have gaudy numbers and pedigrees, but iffy NCAA Tournament prospects. Kentucky has dusted a number of really good teams -- Kansas, Louisville, North Carolina, Texas -- and struggled only rarely, and its perch at No. 1 in KenPom's trusted efficiency-based ratings reflects that. Kansas, meanwhile, lags at No. 11 in KenPom, second in its own state in those rankings, behind No. 8 Wichita State (No. 12 in RPI, for the record).

The chief problem with RPI -- which has this flaw, and others, by design -- is that it weights strength of schedule so heavily. That likely made plenty of sense when RPI was chosen as the Selection Committee's tool in the 1980s, when it was introduced, to reward teams that schedule well and penalize teams that try to game the system by playing against poorer competition.

But institutional advantages help blue-bloods rule RPI year in and year out. Kentucky and Kansas (and teams like them) are able to entice (or pay) good teams to come to Rupp and The Phog, and can afford to schedule neutral-site games worth more in RPI than home ones, without worrying about trying to fill an athletic department's till.

Often, those non-conference showdowns are win-win scenarios: Kentucky didn't just get a win over Kansas, it got the benefit of having Kansas's strength of schedule factor into its own strength of schedule, and even though Kansas lost to Kentucky, it got more from that loss than it could have from virtually other loss it could have taken. (Why is VCU No. 2 in strength of schedule right now? One of its three losses is to Virginia, which is a bigger help than any win on its slate, and another is to Villanova, which functions similarly.)

Being free to construct non-conference schedules with the goal of compiling a strong RPI is a major luxury, and smart scheduling in non-conference play is one of the primary reasons why Duke, North Carolina, Kansas, and Kentucky have topped CBS's pre-NCAA Tournament RPI in 15 of the last 21 seasons, with no other school finishing on top more than once.

And that advantage is available to lesser teams with big budgets, too; the "gaming" of RPI by such teams, often at the expense of mid-majors has long been a criticism of the formula, and there's really no way to fix that without ditching the formula.

The adjustments for home and road wins do help combat that, at least a little. Teams like VCU and Wichita State that have to travel and get creative to get good non-conference games (power conference schools don't want to play at mid-majors, generally, because things like this happen) get extra credit for being road warriors. And teams that lay in wait at home and schedule gimme games for most of non-conference play -- like Kentucky did over an seven-game stretch between its neutral-site bludgeonings of Kansas and UCLA this fall, and as Ohio State famously has under Thad Matta of late -- don't benefit that much from stacking calorie-free wins.

To bring it all back to Kentucky and Kansas, though, those home and road adjustments don't factor into strength of schedule, which relies on raw winning percentage. And opponents' winning percentage is worth double what adjusted winning percentage is, anyway, so Kentucky's 1.000 winning percentage ends up counting more for Kansas than the Wildcats' win over Kansas does for Kentucky.

Is that inane, and maybe a little insane? Sure. But there's no "torturing" of numbers involved here: That's just how RPI works, and always has.

RPI has never been a power rating; it's always been, dating back to its introduction in 1981, about giving the Selection Committee "objective data" to use as an "important aid" in selecting teams for the NCAA Tournament.

Trying to discern the nation's best team by RPI alone, even with its component parts, would be a fool's errand: It tells us only whether a team has won or lost games, the quality of its schedule, and the quality of its opponents' schedules, without factoring in margin of victory, and with only a crude measure of quality of opponent. That Kentucky win over Kansas counts as much for the Wildcats as Florida winning a game over UAB in the Bahamas does for Florida, for example, and though the Kansas and UAB schedules and records get weighed in RPI, that's a really awkward way of trying to discern quality of opponent.

Fortunately, RPI's never been the only factor in selecting teams, and the Committee isn't dumb enough to give Kansas the No. 1 overall seed for its hypothetical No. 1 RPI over an undefeated Kentucky or an undefeated Virginia. Besides, we're still a long way from that being a prospect, and we can attribute some of the irrationality of Kansas over Kentucky to a small sample size: There's plenty of season to go, and plenty of time for numbers to change.

Still: Kansas running the gauntlet that is the rugged Big 12 (with five teams in the top 30 of the NCAA's current RPI other than Kansas) is only going to help its strength of schedule, and Kentucky playing in an SEC that has one team in the RPI top 30 (Arkansas) outside of the Wildcats isn't going to help much. Kansas probably needs to run the table from here until the NCAA Tournament to pull it off, but there's a chance that even an undefeated Kentucky could stay behind Kansas in RPI.

It wouldn't matter to the Selection Committee, which would use the human element of rational judgment to reward the Wildcats (or Virginia) accordingly.

It would, however, spotlight how crazy it is, in 2015, that RPI -- inscrutable enough, despite existing for four decades, that it gets hand-waving rhetorical vitriol, and so flawed that it often values what a program does in setting its schedule than what a team does on the court -- is still critical to college basketball.

We can do better than the freakouts about RPI that don't engage with, assess, and criticize it.

But we can do a lot better than RPI, too.