Ignore The RPI … For Now

For the past several days, I’ve been having a debate with JB34 in the Sports Shack over the RPI and what it means for NC State at this point in the season. It turns out that JB34 is also a writer at StateFansNation.com and today he wrote an article further expounding on his opinions on the matter. Basically, his point is that State’s RPI (100) is bad and that is a cause of alarm.


My counter-argument is that he shouldn’t be looking at the RPI yet. No one should. Yes, I know it’s available and tempting, but that doesn’t mean you should pay it any mind.
For a great explanation of why this is, read the words of Ken Pomeroy (who knows a hell of a lot more about this stuff than I do). Ken wrote a three part essay on the RPI and what it is and is not good for. You can (and should) read the three parts here – one, two, three.
The basic reasoning for why the RPI is something you should ignore is that 75% of the calculation comes from a team’s schedule. Early (or even midway) through a season, these schedules ratings can have huge variances. By the end of the season, as teams regress toward .500 records, their schedules converge to a much narrower gap. By Ken’s measurement the teams with the best and worst schedules over three years varied by an average of only .2140 per year. While it’s 3/4 of your RPI rating, teams don’t have too big of a gap in their schedules, so eventually, a team’s won/loss record comes more into play.
In Ken’s own words:

The range of values in the SOS is also part of the reason the RPI is useless early in the year. SOS has a much greater range this time of year, so it does control the ratings. But as the year progresses, everybody’s SOS gravitates towards .500 and winning percentage becomes more important.

Let’s take the two teams that JB34 used as examples, Clemson and NC State. To a casual observer, NC State is the better team. To the RPI right now though, Clemson is much better. According to Pomeroy’s RPI calculations, Clemson is #39 and NC State is #102. NC State actually has a better record at 11-5 to Clemson’s 10-6, so the difference is obviously in their schedules. If you compare those, you’ll see that Clemson’s opponents have a .5861 record, good for 20th best in the country (this is in large part due to Clemson’s loss to undefeated Boston College). NC State’s opponents on the other hand have a .5151 record, the 125th best in the country. As the season goes on and State and Clemson play more ACC teams, those schedules will converge until they are very similar.
As a contrast, let’s take a look at Pomeroy’s power ratings. This rating system takes into account margin of victory, location of the game (actually, the RPI does this now too, to a degree) and when the game was played. These sorts of power ratings, like Sagarin’s and Mike Greenfield’s, are much more accurate more quickly than the RPI. They really just need a handful of games to get all of the teams connected to be accurate.
According to Pomeroy’s Ratings, Clemson is #49 and NC State is #62. This seems much more reasonable to me. It clearly shows that State has some work to do, but there’s no way that there are 100 better teams in the country (Greenfield has State at #67 and Sagarin has them at #52).
So, forget about the RPI for now. It matters, but it’s not accurate yet. Check back in the last couple of weeks of the season to see where things are. Meanwhile, just hope your team wins. Remember that the RPI is just one of several criteria the selection committee uses and if your team takes care of business and gets to at least 7-9 in the ACC, they’ll probably get a bid.


2 responses to “Ignore The RPI … For Now”

  1. casinos Avatar

    casinos

    casinos These trees unpresbyterated mask’d repassed with care, and formed an appropriate sat-here for the place. In pursuing such a receivi

Leave a Reply

Your email address will not be published. Required fields are marked *