John and Ralph (I think) have pointed out that the normalization
factors in "Yet Another" ranking method are unnecessary.   After
pondering my navel for a couple of seconds,  I guess I have to agree.
Multiplication is associative and commutative regardless of how one
picks the factors:-)   But the rationale for the method is still valid
even if it does reduce to the product of win%,  opp win%,  and opp-opp win%
normalized to the maximum of the opp win% - opp-opp win% products.
 
        John also pointed out that we could just as well scale the top team
to 1.0,  but I think I would prefer to leave the scaling as is since those
values give a *little* bit more information to the reader.
 
        Sorry for getting wound up in engineering type reasoning.
 
        -- Dick Tuthill