Last week, I did a couple posts about the non-con and RPI calculations. I got some e-mails questioning my obsessiveness. I wouldn’t say they questioned my sanity, but they wondered why I was still so bothered by it. Part of it has to do with my biases. My first love in sports is baseball. I’ve always loved the game and I love the numbers. I don’t run my own figures, but I know and understand derivatives, regression charts so that I can get into some of the crazy figures.
Sports like basketball and football are much harder to simply quantify the same way. The RPI, though is an interesting tool that can help provide insight into the quality of a team. And of course, it is a major tool in figuring out the who gets in and the seedings of the NCAA Tourney. It is also one of the few mathematical formulas available in basketball, hence my interest.
It has its flaws, though, and the tweaking done to the formula this year may have been a bit much. Even Joe Lunardi, who has gotten a nice gig at ESPN for RPI and “Bracketology” points out the problems (Insider Subs.)
The NCAA turned the RPI on its ear by weighting road wins (and home losses) at 1.4 while devaluing home wins (and road losses) to 0.6. In other words, winning on the road is now worth more than twice as much as winning at home and losing at home hurts more than twice as much as losing on the road. These values were apparently based on the rather crude arithmetic that, across all levels of Division I men’s basketball, home teams win a little more than two-thirds of the time.
This is part of why Pitt’s overall RPI was hurt so badly. The home losses to Bucknell and Georgetown — providing huge boosts to those team’s RPI really hurt Pitt. Lunardi also fails to mention the other reason for the change — to encourage teams to go out and play some games on the road.
My macro-thinking followed these lines: Every year there are a handful of teams for which the RPI just doesn’t jive with a commonsense appraisal of their respective records.
This is probably to be expected when you’re dealing with more than 300 teams in more than 30 different conferences. If we were to look back at each season, we’d be able to identity these few teams fairly easily and also see that the selection committee did the same in its evaluations. No harm, no foul in other words, provided any new formula didn’t worsen this condition substantially.
All of which brings us to 2005. And, while the new RPI is rewarding most teams that perform well on the road, it is failing miserably in the commonsense department.
Instead of three or four teams each year for which the numbers must be discounted, we have at least triple that (and probably more). Instead of becoming a more reliable tool, I believe the RPI is now much less so.
If you think there will be no outcry, just wait until Selection Sunday. I’m guessing there will be a whole lot of head-scratching as committee members try to figure out how Vermont could lose two out of three and keep an RPI in the high teens.
Lunardi lists a half dozen teams who have RPIs that are just out of whack with common sense. Pitt isn’t one of them, but it helps explain the rankings for Holy Cross and Old Dominion. Teams that are high in the RPI despite a relatively unimpressive record of wins, but lots of “good” road losses.
Next year the Big East schedule will be a mess with 16 teams. TV will drive the schedule, so teams won’t even know who they will play and how many times for at least 4 more months according to Andy Katz.
The Big East expects to let the 16 teams for the 2005-06 season know who they will be playing in June after CBS and ESPN make their selections.
Dates and times wouldn’t be set until over the summer, but the matchups will be dictated by television. So if CBS requests Louisville-Connecticut, you can expect ESPN will want the same game. That means those two teams would be matched up in a home-and-home series.
A number of these scenarios will occur. If both television partners want certain games, then those would likely be the home-and-home matchups. The Big East won’t lock in primary-rival partners like the 12-team ACC does.
The Big East will play 16 games in 2005-06 and go through a two-year cycle with the schedule. Teams will play 10 teams once, three teams twice and two teams not at all.
Expect some games to happen naturally, like Louisville-Cincinnati and Marquette-DePaul.
The ACC sent out its schedule for the next three seasons but then abruptly withdrew it after it was released. The league needs to address equity issues before agreeing to the schedule.
I repeat: the BE will be splitting apart in a bout 5 years.
Add Seth Davis of SI.com to the list of those who have been unimpressed with Chris Taft this year.
Chris Taft can’t possibly still be thinking about the NBA. Can he?
Of further note with Davis. At the end of his column, is a list of weekend games he will be picking on Friday. Davis is now 0-4 in picks involving Pitt. He picked against Pitt, 3 straight times and then went with them against Villanova. He has the UConn game listed. I’ve already sent him a request not to pick Pitt for the rest of the year. If anyone else wants to do so, it can’t hurt. Just be sure to mention your favorite tune from the Radiators.