vpFREE2 Forums

Probability Question for VP short term Risk of Ruin

Nope, this is not another Rob Singer thread.
It is related to the wacky world of online gambling.

Anybody know how to solve this puzzle?

Suppose you are playing dollar 8/6 JOB with $50
You plan to play until you either hit $150 units or
go bust.

The paytable, for the curious, is (for a five coin bet)
2500, 225, 125, 40, 30, 20, 10, 5. But obviously the
nunbers for the RF and SF will have little impact upon

My 2 questions are simple.

1. What is the probability that I will quit by reaching
30 units, instead of busting out?

2. Given the same situation, but with a target of only
$125, what is the probability that I would quit by reaching
my target, instead of busting out?

Also, I'd like to know the general method for calculating
this sort of thing and

Quadzilla

I think you'd have to run a PDF sim.
This is a PDF calculator that does 9/6 JOB:
http://members.aol.com/lotsie/GamblersRuin.html
8/6 JOB would be like 9/6 JOB with -1.1% cashback
alternately, 8/6 JOB is like 8/5 JOB with +1.1% cashback
short pay on the royal will cost in the long term

Note to administrator: Lotspiech changed the web address to his
calculator, link in Bankroll links needs to be updated. That's the
current address above.

--- In vpFREE@yahoogroups.com, "what7do7you7want"
<what7do7you7want@...> wrote:

···

Nope, this is not another Rob Singer thread.
It is related to the wacky world of online gambling.

Anybody know how to solve this puzzle?

Suppose you are playing dollar 8/6 JOB with $50
You plan to play until you either hit $150 units or
go bust.

The paytable, for the curious, is (for a five coin bet)
2500, 225, 125, 40, 30, 20, 10, 5. But obviously the
nunbers for the RF and SF will have little impact upon

My 2 questions are simple.

1. What is the probability that I will quit by reaching
30 units, instead of busting out?

2. Given the same situation, but with a target of only
$125, what is the probability that I would quit by reaching
my target, instead of busting out?

Also, I'd like to know the general method for calculating
this sort of thing and

Quadzilla

I can answer the 2 questions easily with Dunbar's Risk Analyzer for
Video Poker, but the payoffs you listed don't appear complete. Did
you leave out trips? Also, is the RF really just 2500?

--Dunbar

--- In vpFREE@yahoogroups.com, "what7do7you7want"
<what7do7you7want@...> wrote:

···

Nope, this is not another Rob Singer thread.
It is related to the wacky world of online gambling.

Anybody know how to solve this puzzle?

Suppose you are playing dollar 8/6 JOB with $50
You plan to play until you either hit $150 units or
go bust.

The paytable, for the curious, is (for a five coin bet)
2500, 225, 125, 40, 30, 20, 10, 5. But obviously the
nunbers for the RF and SF will have little impact upon

My 2 questions are simple.

1. What is the probability that I will quit by reaching
30 units, instead of busting out?

2. Given the same situation, but with a target of only
$125, what is the probability that I would quit by reaching
my target, instead of busting out?

Also, I'd like to know the general method for calculating
this sort of thing and

Quadzilla

Nope, this is not another Rob Singer thread.
It is related to the wacky world of online gambling.

Anybody know how to solve this puzzle?

Suppose you are playing dollar 8/6 JOB with $50
You plan to play until you either hit $150 units or
go bust.

The paytable, for the curious, is (for a five coin bet)
2500, 225, 125, 40, 30, 20, 10, 5. But obviously the
nunbers for the RF and SF will have little impact upon

My 2 questions are simple.

1. What is the probability that I will quit by reaching
30 units, instead of busting out?

The answer depends on the playing strategy. Do you
also want to find the playing strategy which maximizes
your probability of success?

2. Given the same situation, but with a target of only
$125, what is the probability that I would quit by reaching
my target, instead of busting out?

Also, I'd like to know the general method for calculating
this sort of thing and

Finding the optimal strategy for this type of problem is
extremely difficult, much harder than finding a max-EV
strategy or a min-risk strategy. Why? Because the playing
strategy changes as you get closer to the target.

For example, suppose your target is 30 units and you've
been playing for a while and you're almost there -- you
now have 29 units. Hitting any payoff that gives a net gain
will reach the target, so you should treat all hands that
are better than a high pair as equivalent in value. Then, the
optimal playing strategy will maximize the probability of hitting
a winning payoff before losing the 29th unit. An "equivalent
game" ends up using a payoff of 1 unit for a high pair and
a payoff of 3.2416425 units for all higher payoffs. If you
uses these payoffs in your favorite VP analysis program,
it will give the optimal playing strategy whenever you are
one unit shy of reaching your target. The probability of
reaching the goal before losing the 29th unit is 1/3.2416425.

The optimal strategies with fewer units are more difficult
to compute, and finding the "equivalent value" for each
payoff gets complicated. I've only solved a few cases, and
each one depends on the numbers for all of the cases that
are closer to the target. For a target of 30 units, the optimal
solution will use 29 different playing strategies, one for each
distance from the target.

As you move further from the target, the
playing strategy continues to change until you reach a
point "far enough" from the goal so that the strategy
becomes uniform. This only happens if your distance from
the goal is (much?) more than the number of units paid for a royal.
Out at this distance from the goal, the playing strategy is
min-risk, and the probability of reaching the target is approximately
equal to the probability of playing forever when starting with your
current bankroll. For negative games such as this one, the chance
for success is very tiny unless you start with a very large bankroll.

Bottom line: nobody knows the exact answer to your question. It
is an extremely difficult problem to solve, and I'm quit confident
that nobody has ever produced a precise answer. To solve it exactly
would require a very sophisticated program. If anyone tells you they
have an exact answer, I'd be very skeptical.

···

On Friday 17 March 2006 18:30, what7do7you7want wrote:

Wouldn't the strategy at each point also be dependent on what strategy
would be optimal at other points? It might involve a "cascading"
process. Or are you incorporating that? It sounds like you're
assuming that the strategy at each point can be determined
independently of the strategy at all other points, but I'm not sure if
you're saying that or not.

···

On Friday 17 March 2006 18:30, what7do7you7want wrote:

Nope, this is not another Rob Singer thread.
It is related to the wacky world of online gambling.

Anybody know how to solve this puzzle?

Suppose you are playing dollar 8/6 JOB with $50
You plan to play until you either hit $150 units or
go bust.

The paytable, for the curious, is (for a five coin bet)
2500, 225, 125, 40, 30, 20, 10, 5. But obviously the
nunbers for the RF and SF will have little impact upon

My 2 questions are simple.

1. What is the probability that I will quit by reaching
30 units, instead of busting out?

The answer depends on the playing strategy. Do you
also want to find the playing strategy which maximizes
your probability of success?

2. Given the same situation, but with a target of only
$125, what is the probability that I would quit by reaching
my target, instead of busting out?

Also, I'd like to know the general method for calculating
this sort of thing and

Finding the optimal strategy for this type of problem is
extremely difficult, much harder than finding a max-EV
strategy or a min-risk strategy. Why? Because the playing
strategy changes as you get closer to the target.

For example, suppose your target is 30 units and you've
been playing for a while and you're almost there -- you
now have 29 units. Hitting any payoff that gives a net gain
will reach the target, so you should treat all hands that
are better than a high pair as equivalent in value.

>> Nope, this is not another Rob Singer thread.
>> It is related to the wacky world of online gambling.
>>
>> Anybody know how to solve this puzzle?
>>
>> Suppose you are playing dollar 8/6 JOB with $50
>> You plan to play until you either hit $150 units or
>> go bust.
>>
>> The paytable, for the curious, is (for a five coin bet)
>> 2500, 225, 125, 40, 30, 20, 10, 5. But obviously the
>> nunbers for the RF and SF will have little impact upon
>>
>> My 2 questions are simple.
>>
>> 1. What is the probability that I will quit by reaching
>> 30 units, instead of busting out?
>
>The answer depends on the playing strategy. Do you
>also want to find the playing strategy which maximizes
>your probability of success?
>
>> 2. Given the same situation, but with a target of only
>> $125, what is the probability that I would quit by reaching
>> my target, instead of busting out?
>>
>> Also, I'd like to know the general method for calculating
>> this sort of thing and
>
>Finding the optimal strategy for this type of problem is
>extremely difficult, much harder than finding a max-EV
>strategy or a min-risk strategy. Why? Because the playing
>strategy changes as you get closer to the target.
>
>For example, suppose your target is 30 units and you've
>been playing for a while and you're almost there -- you
>now have 29 units. Hitting any payoff that gives a net gain
>will reach the target, so you should treat all hands that
>are better than a high pair as equivalent in value.

Wouldn't the strategy at each point also be dependent on what strategy
would be optimal at other points?

Yes. I'll use S(k) to mean "the optimal strategy when k units are needed
to reach the target." S(1) is independent of other strategies, but S(2)
depends on S(1) and in general S(n) depends on all those that are
closer to the target.

It might involve a "cascading"
process. Or are you incorporating that? It sounds like you're
assuming that the strategy at each point can be determined
independently of the strategy at all other points, but I'm not sure if
you're saying that or not.

The strategies depend on how many units are needed to reach the goal,
but they do not depend on how many units are in the current bankroll.
So, a player who has 25 units with a target of 30 would play the same
way as a player who has 113 units with a target of 118. The proximity
to the target is what matters, not the proximity to going broke.

···

On Saturday 18 March 2006 09:35, Tom Robertson wrote:

>On Friday 17 March 2006 18:30, what7do7you7want wrote:

Steve, I agree that the answer will depend on the strategy. That
does not make the question intractable for all strategies. The
poster did not specify that he wanted to maximize the chance of
reaching his goal. Many people are content to play a maximum ev
strategy with both a stop-loss and a goal. I assumed that this is
the problem the poster was addressing. For this type of problem, a
quite precise answer CAN be achieved.

Therefore I think your "bottom line" should be changed to "nobody
knows the exact answer to your question IF THE STRATEGY IS OPTIMIZED
TO REACH A SPECIFIED GOAL". For someone happy with the maximum-ev
strategy, a precise answer can indeed by found.

--Dunbar

> Nope, this is not another Rob Singer thread.
> It is related to the wacky world of online gambling.
>
> Anybody know how to solve this puzzle?
>
> Suppose you are playing dollar 8/6 JOB with $50
> You plan to play until you either hit $150 units or
> go bust.
>
> The paytable, for the curious, is (for a five coin bet)
> 2500, 225, 125, 40, 30, 20, 10, 5. But obviously the
> nunbers for the RF and SF will have little impact upon
>
> My 2 questions are simple.
>
> 1. What is the probability that I will quit by reaching
> 30 units, instead of busting out?

The answer depends on the playing strategy. Do you
also want to find the playing strategy which maximizes
your probability of success?

> 2. Given the same situation, but with a target of only
> $125, what is the probability that I would quit by reaching
> my target, instead of busting out?
>
> Also, I'd like to know the general method for calculating
> this sort of thing and

Finding the optimal strategy for this type of problem is
extremely difficult, much harder than finding a max-EV
strategy or a min-risk strategy. Why? Because the playing
strategy changes as you get closer to the target.

For example, suppose your target is 30 units and you've
been playing for a while and you're almost there -- you
now have 29 units. Hitting any payoff that gives a net gain
will reach the target, so you should treat all hands that
are better than a high pair as equivalent in value. Then, the
optimal playing strategy will maximize the probability of hitting
a winning payoff before losing the 29th unit. An "equivalent
game" ends up using a payoff of 1 unit for a high pair and
a payoff of 3.2416425 units for all higher payoffs. If you
uses these payoffs in your favorite VP analysis program,
it will give the optimal playing strategy whenever you are
one unit shy of reaching your target. The probability of
reaching the goal before losing the 29th unit is 1/3.2416425.

The optimal strategies with fewer units are more difficult
to compute, and finding the "equivalent value" for each
payoff gets complicated. I've only solved a few cases, and
each one depends on the numbers for all of the cases that
are closer to the target. For a target of 30 units, the optimal
solution will use 29 different playing strategies, one for each
distance from the target.

As you move further from the target, the
playing strategy continues to change until you reach a
point "far enough" from the goal so that the strategy
becomes uniform. This only happens if your distance from
the goal is (much?) more than the number of units paid for a royal.
Out at this distance from the goal, the playing strategy is
min-risk, and the probability of reaching the target is

approximately

equal to the probability of playing forever when starting with your
current bankroll. For negative games such as this one, the chance
for success is very tiny unless you start with a very large

bankroll.

Bottom line: nobody knows the exact answer to your question. It
is an extremely difficult problem to solve, and I'm quit confident
that nobody has ever produced a precise answer. To solve it

exactly

would require a very sophisticated program. If anyone tells you

they

···

--- In vpFREE@yahoogroups.com, Steve Jacobs <jacobs@...> wrote:

On Friday 17 March 2006 18:30, what7do7you7want wrote:
have an exact answer, I'd be very skeptical.

>For example, suppose your target is 30 units and you've
>been playing for a while and you're almost there -- you
>now have 29 units. Hitting any payoff that gives a net gain
>will reach the target, so you should treat all hands that
>are better than a high pair as equivalent in value.

Wouldn't the strategy at each point also be dependent on what strategy
would be optimal at other points?

Yes. I'll use S(k) to mean "the optimal strategy when k units are needed
to reach the target." S(1) is independent of other strategies, but S(2)
depends on S(1) and in general S(n) depends on all those that are
closer to the target.

I was playing in a blackjack tournament recently, in which the only
thing that mattered was to reach a certain goal within a certain
number of hands. I didn't know how to go about it. I guessed that
generally betting about half of what I needed to reach it was about
right, but I assume that at least on the last hand, betting all of
what I needed would have been right, so maybe the optimal amount to
bet gradually increases as the number of hands remaining decreases.

Steve, I agree that the answer will depend on the strategy. That
does not make the question intractable for all strategies. The
poster did not specify that he wanted to maximize the chance of
reaching his goal. Many people are content to play a maximum ev
strategy with both a stop-loss and a goal. I assumed that this is
the problem the poster was addressing. For this type of problem, a
quite precise answer CAN be achieved.

Agreed. In fact, I think you can get a pretty good approximation by
computing the risk of the given strategy. For this game, I compute
R = 1.00343744 and the probability of turning 10 units into 30 units
is given approximately by:

p = (R^10 - 1) / (R^30 - 1) = 0.32196

So there is about a 32.2% of turning 10 units into 30 units when
using max-EV strategy. Using min-risk strategy gives R=1.00312174
to give a 32.3% chance of success (not much better).

Reducing the target to 25 units gives:

p = (R^10 - 1) / (R^25 - 1) = 0.3897 (about 39.0% chance).

These estimates are overly optimistic, but I don't know how far off
they are. What does your risk analyzer give?

···

On Saturday 18 March 2006 10:10, dunbar_dra wrote:

Therefore I think your "bottom line" should be changed to "nobody
knows the exact answer to your question IF THE STRATEGY IS OPTIMIZED
TO REACH A SPECIFIED GOAL". For someone happy with the maximum-ev
strategy, a precise answer can indeed by found.

--Dunbar

--- In vpFREE@yahoogroups.com, Steve Jacobs <jacobs@...> wrote:
> On Friday 17 March 2006 18:30, what7do7you7want wrote:
> > Nope, this is not another Rob Singer thread.
> > It is related to the wacky world of online gambling.
> >
> > Anybody know how to solve this puzzle?
> >
> > Suppose you are playing dollar 8/6 JOB with $50
> > You plan to play until you either hit $150 units or
> > go bust.
> >
> > The paytable, for the curious, is (for a five coin bet)
> > 2500, 225, 125, 40, 30, 20, 10, 5. But obviously the
> > nunbers for the RF and SF will have little impact upon
> >
> > My 2 questions are simple.
> >
> > 1. What is the probability that I will quit by reaching
> > 30 units, instead of busting out?
>
> The answer depends on the playing strategy. Do you
> also want to find the playing strategy which maximizes
> your probability of success?
>
> > 2. Given the same situation, but with a target of only
> > $125, what is the probability that I would quit by reaching
> > my target, instead of busting out?
> >
> > Also, I'd like to know the general method for calculating
> > this sort of thing and
>
> Finding the optimal strategy for this type of problem is
> extremely difficult, much harder than finding a max-EV
> strategy or a min-risk strategy. Why? Because the playing
> strategy changes as you get closer to the target.
>
> For example, suppose your target is 30 units and you've
> been playing for a while and you're almost there -- you
> now have 29 units. Hitting any payoff that gives a net gain
> will reach the target, so you should treat all hands that
> are better than a high pair as equivalent in value. Then, the
> optimal playing strategy will maximize the probability of hitting
> a winning payoff before losing the 29th unit. An "equivalent
> game" ends up using a payoff of 1 unit for a high pair and
> a payoff of 3.2416425 units for all higher payoffs. If you
> uses these payoffs in your favorite VP analysis program,
> it will give the optimal playing strategy whenever you are
> one unit shy of reaching your target. The probability of
> reaching the goal before losing the 29th unit is 1/3.2416425.
>
> The optimal strategies with fewer units are more difficult
> to compute, and finding the "equivalent value" for each
> payoff gets complicated. I've only solved a few cases, and
> each one depends on the numbers for all of the cases that
> are closer to the target. For a target of 30 units, the optimal
> solution will use 29 different playing strategies, one for each
> distance from the target.
>
> As you move further from the target, the
> playing strategy continues to change until you reach a
> point "far enough" from the goal so that the strategy
> becomes uniform. This only happens if your distance from
> the goal is (much?) more than the number of units paid for a royal.
> Out at this distance from the goal, the playing strategy is
> min-risk, and the probability of reaching the target is

approximately

> equal to the probability of playing forever when starting with your
> current bankroll. For negative games such as this one, the chance
> for success is very tiny unless you start with a very large

bankroll.

> Bottom line: nobody knows the exact answer to your question. It
> is an extremely difficult problem to solve, and I'm quit confident
> that nobody has ever produced a precise answer. To solve it

exactly

> would require a very sophisticated program. If anyone tells you

they

> have an exact answer, I'd be very skeptical.

vpFREE Links: http://members.cox.net/vpfree/Links.htm

Yahoo! Groups Links

Steve, what paytable did you use? The original poster wrote "The
paytable, for the curious, is (for a five coin bet) 2500, 225, 125,
40, 30, 20, 10, 5." But that's missing a payoff, which I'm guessing
is 15 for trips. Also, did you use the 2500 for RF?

--Dunbar

> Steve, I agree that the answer will depend on the strategy. That
> does not make the question intractable for all strategies. The
> poster did not specify that he wanted to maximize the chance of
> reaching his goal. Many people are content to play a maximum ev
> strategy with both a stop-loss and a goal. I assumed that this

is

> the problem the poster was addressing. For this type of

problem, a

> quite precise answer CAN be achieved.

Agreed. In fact, I think you can get a pretty good approximation

by

computing the risk of the given strategy. For this game, I compute
R = 1.00343744 and the probability of turning 10 units into 30

units

is given approximately by:

p = (R^10 - 1) / (R^30 - 1) = 0.32196

So there is about a 32.2% of turning 10 units into 30 units when
using max-EV strategy. Using min-risk strategy gives R=1.00312174
to give a 32.3% chance of success (not much better).

Reducing the target to 25 units gives:

p = (R^10 - 1) / (R^25 - 1) = 0.3897 (about 39.0% chance).

These estimates are overly optimistic, but I don't know how far off
they are. What does your risk analyzer give?

> Therefore I think your "bottom line" should be changed to "nobody
> knows the exact answer to your question IF THE STRATEGY IS

OPTIMIZED

> TO REACH A SPECIFIED GOAL". For someone happy with the maximum-

ev

> strategy, a precise answer can indeed by found.
>
> --Dunbar
>
> > > Nope, this is not another Rob Singer thread.
> > > It is related to the wacky world of online gambling.
> > >
> > > Anybody know how to solve this puzzle?
> > >
> > > Suppose you are playing dollar 8/6 JOB with $50
> > > You plan to play until you either hit $150 units or
> > > go bust.
> > >
> > > The paytable, for the curious, is (for a five coin bet)
> > > 2500, 225, 125, 40, 30, 20, 10, 5. But obviously the
> > > nunbers for the RF and SF will have little impact upon
> > >
> > > My 2 questions are simple.
> > >
> > > 1. What is the probability that I will quit by reaching
> > > 30 units, instead of busting out?
> >
> > The answer depends on the playing strategy. Do you
> > also want to find the playing strategy which maximizes
> > your probability of success?
> >
> > > 2. Given the same situation, but with a target of only
> > > $125, what is the probability that I would quit by reaching
> > > my target, instead of busting out?
> > >
> > > Also, I'd like to know the general method for calculating
> > > this sort of thing and
> >
> > Finding the optimal strategy for this type of problem is
> > extremely difficult, much harder than finding a max-EV
> > strategy or a min-risk strategy. Why? Because the playing
> > strategy changes as you get closer to the target.
> >
> > For example, suppose your target is 30 units and you've
> > been playing for a while and you're almost there -- you
> > now have 29 units. Hitting any payoff that gives a net gain
> > will reach the target, so you should treat all hands that
> > are better than a high pair as equivalent in value. Then, the
> > optimal playing strategy will maximize the probability of

hitting

> > a winning payoff before losing the 29th unit. An "equivalent
> > game" ends up using a payoff of 1 unit for a high pair and
> > a payoff of 3.2416425 units for all higher payoffs. If you
> > uses these payoffs in your favorite VP analysis program,
> > it will give the optimal playing strategy whenever you are
> > one unit shy of reaching your target. The probability of
> > reaching the goal before losing the 29th unit is 1/3.2416425.
> >
> > The optimal strategies with fewer units are more difficult
> > to compute, and finding the "equivalent value" for each
> > payoff gets complicated. I've only solved a few cases, and
> > each one depends on the numbers for all of the cases that
> > are closer to the target. For a target of 30 units, the

optimal

> > solution will use 29 different playing strategies, one for each
> > distance from the target.
> >
> > As you move further from the target, the
> > playing strategy continues to change until you reach a
> > point "far enough" from the goal so that the strategy
> > becomes uniform. This only happens if your distance from
> > the goal is (much?) more than the number of units paid for a

royal.

> > Out at this distance from the goal, the playing strategy is
> > min-risk, and the probability of reaching the target is
>
> approximately
>
> > equal to the probability of playing forever when starting with

your

> > current bankroll. For negative games such as this one, the

chance

> > for success is very tiny unless you start with a very large
>
> bankroll.
>
> > Bottom line: nobody knows the exact answer to your question.

It

> > is an extremely difficult problem to solve, and I'm quit

confident

> > that nobody has ever produced a precise answer. To solve it
>
> exactly
>
> > would require a very sophisticated program. If anyone tells

you

···

--- In vpFREE@yahoogroups.com, Steve Jacobs <jacobs@...> wrote:

On Saturday 18 March 2006 10:10, dunbar_dra wrote:
> --- In vpFREE@yahoogroups.com, Steve Jacobs <jacobs@> wrote:
> > On Friday 17 March 2006 18:30, what7do7you7want wrote:
>
> they
>
> > have an exact answer, I'd be very skeptical.
>
> vpFREE Links: http://members.cox.net/vpfree/Links.htm
>
>
> Yahoo! Groups Links
>
>
>

bet goal/hands
if you lose a hand, bet 3x on next hand, if that hand loses, bet 7x on
next hand ...
adjust for splits/doubles

···

--- In vpFREE@yahoogroups.com, Tom Robertson <thomasrrobertson@...> wrote:

I was playing in a blackjack tournament recently, in which the only
thing that mattered was to reach a certain goal within a certain
number of hands. I didn't know how to go about it. I guessed that
generally betting about half of what I needed to reach it was about
right, but I assume that at least on the last hand, betting all of
what I needed would have been right, so maybe the optimal amount to
bet gradually increases as the number of hands remaining decreases.

> Steve, I agree that the answer will depend on the strategy. That
> does not make the question intractable for all strategies. The
> poster did not specify that he wanted to maximize the chance of
> reaching his goal. Many people are content to play a maximum ev
> strategy with both a stop-loss and a goal. I assumed that this

is

> the problem the poster was addressing. For this type of

problem, a

> quite precise answer CAN be achieved.

Agreed. In fact, I think you can get a pretty good approximation

by

computing the risk of the given strategy. For this game, I compute
R = 1.00343744 and the probability of turning 10 units into 30

units

is given approximately by:

p = (R^10 - 1) / (R^30 - 1) = 0.32196

So there is about a 32.2% of turning 10 units into 30 units when
using max-EV strategy. Using min-risk strategy gives R=1.00312174
to give a 32.3% chance of success (not much better).

Reducing the target to 25 units gives:

p = (R^10 - 1) / (R^25 - 1) = 0.3897 (about 39.0% chance).

These estimates are overly optimistic, but I don't know how far off
they are. What does your risk analyzer give?

Steve, I get values that are very close to 1/2 the values you have.

I'm assuming this is the paytable the original poster intended:
2500,250,125,40,30,20,15,10,5,0. Using those payoffs, I got the
hand frequencies from WinPoker. Then I used Dunbar's Risk Analyzer
for Video Poker to answer the Goal vs Ruin questions.

I ran 100,000 trials in DRA-VP until each trial ended in ruin. DRA-
VP keeps track of whether the Goal was reached before ruin. Here
are the results for a 10-unit bankroll:

Goal Success
(units) Rate
30 16%
25 19%

It's interesting that my figs are so close to 1/2 your figs.
Coincidence, or something to do with trip risk vs end-point calcs?

--Dunbar

···

--- In vpFREE@yahoogroups.com, Steve Jacobs <jacobs@...> wrote:

On Saturday 18 March 2006 10:10, dunbar_dra wrote:

> > Steve, I agree that the answer will depend on the strategy. That
> > does not make the question intractable for all strategies. The
> > poster did not specify that he wanted to maximize the chance of
> > reaching his goal. Many people are content to play a maximum ev
> > strategy with both a stop-loss and a goal. I assumed that this

is

> > the problem the poster was addressing. For this type of

problem, a

> > quite precise answer CAN be achieved.
>
> Agreed. In fact, I think you can get a pretty good approximation

by

> computing the risk of the given strategy. For this game, I compute
> R = 1.00343744 and the probability of turning 10 units into 30

units

> is given approximately by:
>
> p = (R^10 - 1) / (R^30 - 1) = 0.32196
>
> So there is about a 32.2% of turning 10 units into 30 units when
> using max-EV strategy. Using min-risk strategy gives R=1.00312174
> to give a 32.3% chance of success (not much better).
>
> Reducing the target to 25 units gives:
>
> p = (R^10 - 1) / (R^25 - 1) = 0.3897 (about 39.0% chance).
>
> These estimates are overly optimistic, but I don't know how far off
> they are. What does your risk analyzer give?

Steve, I get values that are very close to 1/2 the values you have.

I'm assuming this is the paytable the original poster intended:
2500,250,125,40,30,20,15,10,5,0. Using those payoffs, I got the
hand frequencies from WinPoker. Then I used Dunbar's Risk Analyzer
for Video Poker to answer the Goal vs Ruin questions.

The original poster gave a payoff of 225 for straight-flush, but other
than that you've used the payoffs that I used.

I ran 100,000 trials in DRA-VP until each trial ended in ruin. DRA-
VP keeps track of whether the Goal was reached before ruin. Here
are the results for a 10-unit bankroll:

Goal Success
(units) Rate
30 16%
25 19%

It's interesting that my figs are so close to 1/2 your figs.
Coincidence, or something to do with trip risk vs end-point calcs?

That is very interesting. I thought the RoR estimate would be much
closer than that.

···

On Saturday 18 March 2006 12:34, dunbar_dra wrote:

--- In vpFREE@yahoogroups.com, Steve Jacobs <jacobs@...> wrote:
> On Saturday 18 March 2006 10:10, dunbar_dra wrote: