vpFREE2 Forums

MCR vs MER

I noticed Cindy's tool doesn't go below 1%, but that can be solved
with this formula:

Bankroll@ror1=Bankroll@ror2 x ln(ror1)/ln(ror2)

If you can "supplement" your own play (I hear you can sell casino
chips on ebay for more than their face value, if not that you can
always be a strip gigolo, or set up a meth lab in your hotel room) to
the tune of 0.5% cashback, using Cindy's tool, the bankroll@1%ror is:
3964 bets. ln(.001)/ln(.01)=1.5, so the 0.1%ror (99.9% survivability)
bankroll is 5964 bets ($7,432.50 on five coin quarters).

But all you really have to remember is this:
kelly=approximately variance/(ER+cashback-1) bets
bankroll under kelly/3= expect to bust out
bankroll at kelly/2= expect to break even (optimal strategy is minrisk)
bankroll at kelly= expect max bankroll growth (optimal strategy is
between minrisk and maxER)
bankroll over kelly= less risk but less bankroll growth (optimal
strategy approaches maxER)

This is all just math, it doesn't require faith to make it valid, nor
does disbelief or ignorance make it invalid.

--- In vpFREE@yahoogroups.com, "nightoftheiguana2000"
<nightoftheiguana2000@...> wrote:

···

--- In vpFREE@yahoogroups.com, "Harry Porter" <harry.porter@> wrote:
> But that doesn't seem unrealistic at all (the funding part, not the 5K
> hands, day in and day out ;). I can well see someone starting out
> with an upfront $10K loss allowance and being willing to supplement
> that with as much as $3K-$5K per year, unceasingly, as an
> entertainment budget against 1500+ hrs. of play a year. I expect that
> if these numbers need be adjusted to a moderate degree to get to a
> 99.9% survival probability that I'd be quite comfortable with that as
> well.

Your numbers look about correct, for $1.25/hand. I suspect the average
quarter player would be surprised at their size however.

--- In vpFREE@yahoogroups.com, "nightoftheiguana2000" <nightoftheiguana2000@...>
wrote:

This is all just math, it doesn't require faith to make it valid, nor
does disbelief or ignorance make it invalid.

(1) We'll in some way it does require some faith or something other than math. Indeed
Kelly only makes rigorous mathematical sense if the follower has the "correct" utility
function. Or turned around the other way, if you happened to be a person who has certain
utility function then (and only then) kelly betting makes sense for you. If you don't happen
to possess such a utility function, then, well, strictly speeking, out-of-the-box kelly is
NOT for you. It has already been shown that most people, btw, don't have have (follow ?
believe in ? posess?) the correct utility function for strict kelly. Hence, when "kelly" is
followed, it is usually done so in a "modified form".

(2) While the concepts behind computing Kelly and RoR in advance of play are useful
figuring out what game to play (or what game not to play), they are pretty useless during
actual gambling and shouldn't (in general) effect game play choices after play begins. But
they can be "extended": one uses an instantaneous RoR , and instantaneous Kelly that
takes into accoung one's real-time results. For example, one can compute the instaneous
RoR (finite # of hands or infinite) after evey hand of VP -- and determine if one should
"quit" or "continue" based on a some arbitray RoR threshold. Same goes for kelly. Based
on current result one can compute the correct bet and stratgey to follow. BUT, one should
compute the kelly stuff first, since the RoR depends on the strategy (which affects the
PDF). But kelly depends on ones utility function.. which tends to change as gambling
proceeds (Few of us really know how we will feel after a bug loss until we have a big loss)
Oh well... its an awful lot of computation to do infront a VP machine--- I guess one could
bring vast tables of precomputed results along. LOL.

(3) Being ignorant of one's utility function doesn't make the math of "kelly" invalid per se,
but rather makes the adherence to strict kelly inapproraite and, well, wrong, for that
person. Does anyone really have a log utility function ? Maybe.. but likely not.