I've been following some of the threads about Quinnipiac versus the
civilized world and the various heated arguments that invoke the RPI and its
close cousin, the PWR. These discussions have inspired me to finally ask a
question that I've wondered about for some time.
 
Given my background of a PhD in Theoretical Physics and a strong
background in mathematics and probability and statistics, it has always seemed
self evident to me that rating algorithms like the KRACH or the CHODR (or its
close cousin, the CCHP), although not perfect, do make a great deal of
sense and have a firm basis in probability and statistics.
 
The RPI (the rating one; not the Rensselaer one), on the other hand, has
never made any sense to me. I understand its general motivation, of trying
to take some account of differences in strength of schedule, but its
shortcomings and failure modes are so pitifully obvious, I can't possibly
imagine how somebody conjured it up in the first place. I can only imagine
that it was invented by some guys who had their degrees in Phys Ed and never
passed a math course in their life.
 
(I do realize that if all teams' strengths of schedule are not too
dissimilar, then any not-totally-insane rating system will give a
not-too-unreasonable result, so even the RPI won't look too bad. But for cases where
strength of schedule makes a significant difference, as in the great MAAC
controversy, the RPI fails miserably.)
 
So that leads to my question: who *did* invent the RPI, how did they
choose the particular formula that they did, and why would anyone ever believe
that it makes any sense?
 
Bob "Have I offended enough people?" Stagat
RPI (the Rensselaer one; not the rating one) '64  '68
 
HOCKEY-L is for discussion of college ice hockey;  send information to
[log in to unmask], The College Hockey Information List.