I've started to do some "first-principle" computations (not Monte
Carlo
simulations) regarding Rob's "strategies". I think discussion of
these
computations (or similar ones by others) would be an excellent
subject
for this group-- so long as we do stay away from making claims we
can't
substantiate or comments about Rob himself.
My computations are based on a kind of Bayesian-like approach using
exact (or nearly exact) PDFs and finite hand RoR curves. Here is how
I
sum up RS's strategy: You start with a session bank roll (or stake)
that is much larger than your win goal. If the win goal is achieved
you stop playing. If you lose your stake you stop playing. You chose
the game (bet amount, etc) based on how much you have already
lost/won
this session and some time factor. [Is this description correct
enough?]
At least to first order (neglecting the change of game and time
constraints), the RS strategy boils down to assuming that P(LA;SA),
the
probability of losing some LARGE amount (LA) before wining at least a
small amount (SA) is small enough that P(LA;SA) << SA/LA. Hence,
one
could assume (in the long run) that someone following this strategy
expects to win the SA many times (in separate successful sessions)
before losing the LA once, at least most of the time.
Is this a reasonable way to approach a mathematical description of RS
strategy? Comments?
--- In vpFREE@yahoogroups.com, "Harry Porter" <harry.porter@v...>
wrote:
···
martin stern wrote:
> maybe i have been lucky, but the highest i highest
> machine i progressed to using singers method was the
> $5.00 machine...
My aplogies for my comment ... I took you statement to suggest that
you had never had cause to play other than $.25/$.50 machines.
- H.