I think that we should not lose sight of one very important facet
of this discussion.   What we all are seeking to discover is how to level
the playing field so that it is fair to all teams.
 
        If all teams played a true interlocking schedule,  then of course
win% would be the only thing that counted.   If all teams played perhaps not
the *same* schedule,  but schedules of equal strength,  then win% would still
be the only thing that counted.   And,  of course,  when the schedule
strengths are unequal,  then somehow we must find a method to correct the
win% so that meaningful comparisons can be made.   How to do this?
 
        I've been giving this some thought,  and have not yet got my arms
around it  --  but I know what we shouldn't be doing.   We should not be
drawing any conclusions from the amount of variation produced by win% or by
strength of schedule.   We simply don't have an analysis yet which quantifies
the effect of SOS on win%.   Consider Team A which has a win% of 0.600 against
opp% of 0.450.   Is that better or worse than Team B which has a win% of 0.400
against opp% of 0.550?   We simply don't know,  and so far the discussion has
boiled down to intuition.
 
 
        In proposing YAM2,  for instance,  I call it an intuitively based
method.   I would be quite happy to dump YAM2 for something more rigorously
derived if we had an analysis which justified it.   What we need is an eager
MS student to grab this problem and do a thesis on it.   I can think of no
better win-win situation than to be able to tell my professor that I was
going to the hockey game to work on my research!!!!   In any event,  the data
are just sitting there waiting for someone to try.:-):-)
 
        -- Dick Tuthill