Monday, February 16, 2009

PWR Update - Case of the Mondays

I, unlike every other soul in Washington DC, had to go into work for President's Day. I did, however, notice something interesting in the PWR. The most current version still has UW at #15, but with a slightly different line. Sunday's update had this:



Record RPI
Rk Team PWR Rk W-L-T Win % Rk RPI

15 Wisconsin 10 17t 16-11-3 .5833 15 .5380

Monday brings us this:



Record RPI
Rk Team PWR Rk W-L-T Win % Rk RPI

15 Wisconsin 10 17t 16-11-3 .5833 15 .5381

What team(s) do we have to thank for that minute increase in RPI? None other than Bemidji State and Robert Morris, as they were the only teams playing on Sunday, thus they represent the only new data inputs into the system. Their 3-3 tie was probably a pretty good game, with RM overcoming a 2-goal third period deficit with back to back goals only 38 seconds apart.

I bring this up for two reasons:

It shows how fickle the PWR can be. A single game between two teams that UW hasn't played changes the data. This alone isn't all that remarkable, but...

The margin between teams is incredibly small. Minnesota is winning the individual comparison with UW thanks to their edge in RPI. Minnesota's RPI sits at .5384. The increase from the BSU-RMU tie puts Wisconsin at .5381, just .0003 behind the rodents. That's three ten-thousandths of a point.

So what? A small change in the data produces a small change in the results.

However, if Wisconsin somehow makes up those three ten-thousandths of a point in the RPI, they then win the comparison with Minnesota, flipping it in their favor, and bumping UW up one spot in the rankings - the coveted 14th position.

The margin is razor thin.