vpFREE2 Forums

Bob Dancer's LVA - 19 MAY 2015

#1

nightoftheiguana2000
        <mailto:nightoftheiguana2…@…com?subject=Re%3A%20Bob%20Dancer%27s%20LVA%20-%2019%20MAY%202015>
        wrote:

"The Jazbo/Sorokin formula is nice, I use it a lot, but there are three
problems with the Jazbo/Sorokin formula:
1. It assumes play forever. For the non-infinite human timespan you need
something like Dunbar's Risk Analyzer.
2. What do you do if any risk of bankruptcy is unacceptable? In other
words, what if your desired ROR is 0%? Other than an infinite bankroll,
Jazbo/Sorokin has no answer to that question, Kelly does. Kelly answers
the question of how do you play without going bankrupt and on a finite
bankroll.
3. What is an optimal betsize or in the case of Jazbo/Sorokin, what is
the optimal level of risk? Jazbo/Sorokin has no answer to this question,
Kelly does."

In response:

1. Sorokin/Jazbo does not necessarily assume playing forever. Optimum
Video Poker also uses it to calculate Risk of Ruin Before a Royal Flush.

2. What to do if any RoR is unacceptable? Kelly does not answer this
adequately for video poker because it is impossible to vary your bet
size in small increments. The only way to absolutely avoid ruin is,
DON'T PLAY AT ALL.

3. What is optimum (or optimal if you prefer that word) is highly
subjective. My book, Video Poker Optimum Play, provides tools to assist
each player in answering this question for himself. Some of these tools
are implemented in Optimum Video Poker.

How to best use Sorokin/Jazbo? Select a game of your choice and try
Sorokin with your current bankroll and each of the available machine
denominations, then pick the denomination that is most compatible with
your emotional tolerance of risk.

[Non-text portions of this message have been removed]

#2

Dan wrote: "2. What to do if any RoR is unacceptable? Kelly does not answer this
adequately for video poker because it is impossible to vary your bet
size in small increments. The only way to absolutely avoid ruin is,
DON'T PLAY AT ALL."

If this were true, Kelly would be useless in the real world. When was the last time you were able to bet $23.42 at a blackjack table? The solution is that it's not true, it's just more Kelly bashing.

Now, I'll let you in on a little secret. Any fractional betting system will avoid bankruptcy. It's just that Kelly is the optimal fractional betting system if you assume a log utility function, or in plain English it is important to you how long it takes to double your bankroll. And yes in the real world there are only certain fixed bet sizes, but given this constraint Kelly is still the optimal solution given a log utility function.

Here it is 2015 and it still amazes me the kind of misinformation that floats around about Kelly. Seriously, if you want a basic understanding of Kelly that doesn't require any math background, read Poundstone's "Fortune's Formula".

[Non-text portions of this message have been removed]

#3

At your recommendation, I read this and it was interesting, but I
think it exaggerated the value of the Kelly Criterion. I believe
Thorp's success was primarily due to positive expected value, not that
he didn't overbet his bankroll. "Sloppily" applying the Kelly
Criterion, without significantly and consistently overbetting, with
significant positive expected value, will still accomplish almost all
of what Thorp did.

···

if you want a basic understanding of Kelly that doesn't require any math background, read Poundstone's "Fortune's Formula".

#4

007 wrote: ""Sloppily" applying the Kelly
Criterion, without significantly and consistently overbetting, with
significant positive expected value, will still accomplish almost all
of what Thorp did."

OK, but you still need a definition for "significant and consistent overbetting". The book also has examples of those who forgot that step.

There are two hurdles here, the first is to find an edge, the second is to not overbet your bankroll. You have to clear both if your expectation is bankroll growth with zero risk of ruin.

[Non-text portions of this message have been removed]

#5

I wrote: "OK, but you still need a definition for "significant and consistent overbetting"."

And I'm proposing: keep EV-Variance/Bankroll/2 positive. If you have a monster bankroll, this just reduces to keep EV positive or keep yourself from running out before you die, but most of us don't have monster bankrolls. If you have tolerance of risk of ruin and a gameplan for potential bankruptcy, then you can be more aggressive and use the ROR equations and software.

[Non-text portions of this message have been removed]

#6

Or another, maybe better way to put it:

If your goal is bankroll growth with zero risk of ruin, keep your bankroll over variance/edge/2 bets (or half the Kelly number if you know it).

[Non-text portions of this message have been removed]

#7

noti said: "keep yourself from running out before you die, but most of us don't have monster bankrolls."

I don't mean to argue. I just mean to bring up my own personal circumstance (and maybe the circumstance for a lot of "recreational" players). It is not that I have a "monster bankroll". It is just that I play in such a way that I have my bankroll replenished faster than it is drained. My "gambling bankroll" is a line item in my entertainment budget.

[Non-text portions of this message have been removed]

#8

noti said: "keep yourself from running out before you die, but most of us don't have monster bankrolls."

I don't mean to argue. I just mean to bring up my own personal circumstance (and maybe the circumstance for a lot of "recreational" players). It is not that I have a "monster bankroll". It is just that I play in such a way that I have my bankroll replenished faster than it is drained. My "gambling bankroll" is a line item in my entertainment budget.

That's one of the problems with the Kelly Criterion, since it assumes
no outside source of income. Going by one's "gut" has its dangers,
too, but most people who lack a sophisticated knowledge of the Kelly
Criterion, including me at least most of my life, have a reasonably
accurate idea of what overbetting means.

#9

007 wrote: "That's one of the problems with the Kelly Criterion, since it assumes
no outside source of income."

Other sources of income, or for that matter taking funds from the bankroll, can be modeled as changes to EV.

I already provided an example which was $5 double double I believe, which has a negative Kelly CE. As I explained, this cost to play has to come from somewhere. One possible source would be your day job, in effect, if you add from your day job to your bankroll at the rate of the negative CE, you would be at bankroll neutral. And of course if you add to your bankroll faster than it is depleted, you will grow your bankroll, nothing wrong with any of that.

007 wrote: "Going by one's "gut" has its dangers,
too, but most people who lack a sophisticated knowledge of the Kelly
Criterion, including me at least most of my life, have a reasonably
accurate idea of what overbetting means."

Isn't that just a survivor bias? You don't know of anyone who went broke from overbetting?

[Non-text portions of this message have been removed]