My sim has a windows interface (duh) with radio buttons for {
Singles, Doubles, Neutral}.
Now consider that Dania has 32 singles games per week, that would
make approximately 1664 per year.
To represent an entire year of ALL Dania singles, then, I could
set the # of games simulated to 1664. As opposed to BRNE's 1 billion
superfecta run. Doesn't that tell you something? It should.
Now, let's say that we choose to use the singles option with the
server bias such that we expect the server to win 53% of all points.
We also inspect the outcome of key tri pairs such as 841 / 842,
851 / 852, 741 / 742, and so on.
It turns out that owing to the small sample size of 1664 games,
there is wild disparity from run to run. One time I will get 841
= 7 / 842 = 14, the next it will be 13 to 5. There is similar
variation for other "runout sensitive" numbers.
So again, I ask Tiger, "If I gave you the results for a batch
of singles games (here, 1664), could you tell me with confidence
how many were runouts?
If you concede that 1 season is not enough, then how many do you
need?
