I find it interesting that someone else who is taking an analytical
approach would independently reach the same approximate conclusion.
Your 9-5 top choice corresponds to my "about 1 in 3"
and the 18-1 at the other end matches my "1 in 20" limit
very closely.
I think you can go outside that range at times - eg: Alberdi and
Cuvet won 8 out of 16 games in the 2000 NAJF Qualifying Tournament
(imagine their results in posts 1-2-3) - but overall it is not
practical.
