I'm working on setting up a simulation that will measure ROR on multi-
play games over a limited time frame such as 200K hands, rather than
assuming eternity as the standard ROR formulas do. First, are there
any formulas that will give me expected results from such a
simulation or checkpoints along the way? Second, does anyone see any
labor-saving or resource-saving modifications I could make to the
procedure described below?
This is a much more difficult simulation than for a single-line game
because of the co-variance between hands. To simulate a single-line
game all you need is the table of finished-hand probabilities, which
can easily be derived from the paytable and a game analysis that
shows how much of the ER comes from each type of finished hand. You
don't care what's dealt and drawn, only the final result. Apply a RNG
and a tracking method and, voila, you can do any number of
simulations of any length.
For the multi-play, I'm thinking I have to start with the 2598K
unique poker hands (actually I'll use a table of 134K suit-
independent possibilities along with a column that defines the
frequency of each in the 2598K universe) and apply the RNG there,
then apply a second RNG to obtain each hand's results and tabulate
from there. Not difficult in concept, and I already have the 134K
table built for JOB-based games, but it's going to be a
memory/resource hog and relatively slow to run. Any ideas to speed
things up?