The best way to prove this is to take a normal run (say 1 million
times) and flag each runout by each post and log it's associated
combo. So you will see more runouts as we approach 678.
Then, run another simulation (same # of times) with a server bias
and flag each combo the same way. Compare the two and theoretically
you will see more 52's that 25's or 47's than 74's than you would
without a server bias.
Regarding the stopping at game point scenario, the reverse of
that is not true. A square is a rhombus, but a rhombus is not
a square. Or is it trapezoid....
