Rating the Ratings
Call me a geek, but I like computers. I’m not into the AGP graphics cards or the Flash SRAM memory. Nor am I into chat rooms and role-playing games. The Internet, though, is a pretty damn cool thing. Heck, without the Internet, you wouldn’t be reading my random thoughts. And I think we’d all agree that the world is a better place now that Ermitage has a column.
While computers have changed the world and how we work (I think I read that in the first sentence of some Computers 101 book), it has also greatly affected college athletics. Most would agree it has ruined college football and enhanced college hoops. Instead of playing the “who-beat-who” game in our living room, we actually can figure it out mathematically and display it worldwide just like Jerry Palm and his collegerpi.com. Or just like Jeff Sagarin and his widely published rankings. These are exciting times. But how do we know which rankings to trust and which to ignore? Let me be your geeky, computer-loving tour guide of the college basketball rankings.
Human Polls – Before I review the computer polls available, let me first turn my attention to the much-beloved AP, ESPN/Coaches and Hoopville polls. Every year I am amazed by how important it is to people that their team is ranked in the human polls. Even in November, there is careful analysis by fans of every team about where they ended up that week. And it isn’t just fans of unheralded teams looking for respect. It’s also the Duke fans of the world fretting over why they’re fourth when they should be second. To me, rankings of all kind are virtually useless before February. I cannot believe the heartache people go through surrounding these early-season popularity contests.
AP poll – The AP poll, to me, is the Don Perignon of human polls. This shouldn’t surprise you seeing as how I’m a journalist. I believe the voters in the AP poll take the job more seriously than the other human polls, and the sheer volume of votes available makes the poll legit. Sure, there are some regional biases. And sportswriters generally only see the team they cover and the teams they play. However, sportswriters are often addicted to sports and I feel that when they’re not covering sports, they’re watching them. God bless their wives.
Coaches poll – Coaches are stubborn people. And loyal people. They’re also busy people. And for these three reasons – they make bad voters. While sportswriters will usually err on the side of pessimism (“this team sucks” and “this conference sucks” or “this catered food is terrible” are things often heard in a press box), coaches will err on the side of optimism (“I think our conference is the best,” or “Our conference leader can play with anybody in the country” are things heard in a basketball office). Therefore, there is a heavy amount of favoritism in the Coaches poll. Not to mention that some of the coaches just pass the thing to a basketball manager to fill out, defeating the purpose of getting the coaches opinion.
Hoopville Top 25 Poll – I honestly think the Hoopville poll is a good one. (Ed: Gee, thanks for the endorsement, Mike!) The writers here are basketball junkies and generally do not play favorites. The only drawback is that Hoopville is not yet large enough to have a credible poll. The whims of 30-plus writers, even the best ones, should not decide who are the best teams.
RPI – The centerpiece of water cooler discussions in March is considered the primary indicator of how the selection committee will seed a team. Its formula is simple – 25% team winning percentage (WP), 50% opponents’ average winning percentage (OWP), and 25% opponents’ opponents’ average winning percentage (OOWP). Got that?
The NCAA, of course, maintains that it adds some special quirks to the formula to get the final ratings. There is no real good answer on what the NCAA’s super-special, double-secret probation quirks are – except that they’re designed to further inflate the ego of the NCAA. In addition to the RPI, there’s the adjusted RPI, which removes RPI results that are not significant (namely, wins that decrease a team’s rating and losses that increase the rating). This RPI is designed to show which teams have inflated rankings, and which are potentially better than their current ranking. In my opinion, the RPI rankings are a solid statistical model for ranking the teams. It does have its deficiencies, in that it doesn’t account for where the game is played or for the margin of victory. I do find, however, that as the season wears on that it most accurately reflects the power structure of college basketball. Jerry Palm at www.collegerpi.com runs the best site for this info, while Ken Pomeroy also has a good site.
Pomeroy Rankings – Speaking of Ken Pomeroy, he has his own rankings that he’s modestly named after himself. Unlike the RPI, his rankings take into account the venue in which the game was played and the margin of victory. The purpose of his rankings is to get a predictive score of a potential game, which you can arrive at by taking the ranking of the higher rated team and subtracting the ranking of the lower rated team. Perhaps more useful in gambling than seeding teams, the ratings are interesting, but seem to favor teams that run up big wins against a weak schedule.
Sagarin Rankings – Another predictive ranking system is Jeff Sagarin’s model. Another modest fellow, Sagarin has been posting his rankings in the USA Today since 1985. His rankings, however, seem to put more emphasis on who you’ve played. The Sagarin rakings also take into account the margin of victory and the venue. But he also posts his rankings eliminating those two factors. Again, BCS teams that routinely handle smaller schools are favored here, but it is an informative and polished rating system.
There are certainly other ranking systems out there, including Mike Greenfield’s Team Rankings. It seems that any dude with a theory and a computer can put these things out there. Enter Will O’Hargan, a Ball State student, who put together the ORI rankings, which is a statistical model that just averages the four computer polls (RPI, Sagarin, Pomeroy and Greenfield) with the human polls. My guess is that the NCAA is probably a lot like O’Hargan come selection time, averaging each of these efforts to get an idea of where teams really stand.
Luckily for college basketball fans, it doesn’t matter how many computers you enter into the mix because the championship is always played for on the court. These rankings are like ammunition in your belt when you’re firing away about how great your old alma mater is at hoops. Or maybe they’re just like a microbrew offering at the stadium – a nice but not necessary option that just makes the whole experience better. Fire away and drink up – that’s exactly what college basketball is all about.
