December 30, 2002 (edited Jan. 15, 2003)

Given my tendency to dwell on the negatives of NCAA selection processes, it is only fair to give the NCAA credit when they take a step in the right direction. Most of the press this year has gone to the expansion of the D-I college hockey tournament from 12 to 16 teams, but several other changes have been made that deserve notice.

First and foremost, the RPI formula has been changed to give more weight to schedule and less weight to record. Under the old system, the college hockey RPI was among the worst rating systems in all of college sports. If a mathematically correct system is defined by:

``  rating = record + schedule,``

the old RPI was equal to record + 0.45 * schedule. This means that the RPI rating was not a fair estimate of the team's strength, but was biased in favor of teams with easier schedules. Thus one could "beat the system" by finding the easiest non-conference opponents possible. Under the new system, it is roughly record + 0.70 * schedule, which is still not perfect but is a huge improvement. The current system resembles the biases in the college hockey polls very well.

The second change was that the pairwise ratings no longer consider a team's record in the last 16 games. While it is natural to want "hot" teams in the tournament, the rule as it stood helped out teams whose last 16 games were against weak opponents.

A final change is that the pairwise comparisons are now made against other teams with RPIs of 0.500 or higher rather than with records of 0.500 or higher. This is significant, since a team in a weak conference was able to pad its pairwise ranking with wins over weak teams with winning records. This is no longer the case.

The bottom line from the two changes is that it should be harder for teams to sneak into the tournament by playing easy schedules. Interestingly, this probably won't hurt the CHA and MAAC teams, as they were already treated "specially" by the committee. What will change is that the new system should more correctly account for the superiority of WCHA and Hockey East over CCHA and ECAC.

Note: if you use any of the facts, equations, or mathematical principles introduced here, you must give me credit.