vpFREE2 Forums

Bob Dancer's CasinoGaming Column - 3 NOV 2009

The Kelly system can work fine for games where you can make
relatively small changes in your bets, allowing you to bet a
calculated percentage of your bankroll, but that is rarely possible
in video poker. The only situation I can think of would be a 100-play
machine that lets you play any number of hands.

Also, Kelly is not very good for games with widely skewed
probabilities and payoffs, including video poker.

The only reliable calculation for risk is the Sorokin formula, first
applied to video poker by (I think) jazbo, and included as a feature
in Optimum Video Poker.

Dan

···

At 9:41 PM +0000 11/4/09, nightoftheiguana2000 wrote:

It's all about the size of your bankroll. Which is why the Kelly
system is so important. The Kelly system produces the maximum
average bankroll growth. The rich get richer, the poor get poorer,
and the middle steadily disappears.

--
Dan Paymar
Author of best selling book, "Video Poker - Optimum Play"
Developer of VP analysis/trainer software "Optimum Video Poker"
Visit my web site at www.OptimumPlay.com

"Chance favors the prepared mind." -- Louis Pasteur

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

The Kelly system can work fine for games where you can make
relatively small changes in your bets, allowing you to bet a
calculated percentage of your bankroll, but that is rarely possible
in video poker. The only situation I can think of would be a 100-play
machine that lets you play any number of hands.

The Kelly system still works, even when you can't make that optimum bet of say $7.457999 . Given the bets that you can make, the Kelly system tells you which one is optimal.

Also, Kelly is not very good for games with widely skewed
probabilities and payoffs, including video poker.

As long as you have the correct formula, it is. An approximation for video poker like games is edge/variance.

The only reliable calculation for risk is the Sorokin formula, first
applied to video poker by (I think) jazbo, and included as a feature
in Optimum Video Poker.

That's a good formula, but it only tells you the chance of losing your entire bankroll on a given gamble. The Kelly system tells you how to achieve maximum average bankroll growth. The approximate Kelly bankroll (variance/edge bets) generally corresponds to a Sorokin risk of ruin of about 10%, and the double Kelly bankroll generally corresponds to a Sorokin risk of ruin of about 1%, but the Kelly system is much more than just a bankroll number. It is a system, the only system that maximizes average bankroll growth. Is average bankroll growth important? According to that last Dancer article it is (if you want to continue to play video poker in the future). Dancer consults to the casinos, so his predictions are quick likely to come true.

···

--- In vpFREE@yahoogroups.com, Dan Paymar <Dan@...> wrote:

The risk formula which you and NOTI are refering to as the "Sorokin formula" would be better named the "jazbo/Sorokin formula". Since I am at least partially responsible for the mis-naming of the formula, I want to jump in here to try to set it straight.

The formula itself has been around for a long time. But jazbo was the first to recognize that it could be applied to video poker: jazbo sent a note to his listserv group about it in December of 1998.

A couple of months later, a Russian mathematics student, Evgeny Sorokin, posted a cryptic RoR question about a 3-payoff game on the old bjmath.com site. He answered the question himself a short time later. His posts languished for several weeks until I realized what he meant, and more importantly, realized that his solution could be applied to video poker.

I posted my "Eureka" moment on bjmath, and MathBoy quickly realized that cashback could be included. We wrote an article titled "Risk Of Ruin for Video Poker and Other Skewed-Up Games" that was published in the Fall 1999 (Vol XIX, #3) issue of "Blackjack Forum".

Neither MathBoy nor I had an inkling of jazbo's earlier contribution until after the article was written.* Nor did any of the 3-4 people to whom we sent a draft. That's understandable--we were all blackjack players who had only recently begun to even think about video poker.

MathBoy and I did not refer to the formula as the "Sorokin formula" in our article. We ackowledged Sorokin's contribution, but referred to the formula as the "generalized risk equation". However, I think it's appropriate to call its application to video poker the "jazbo/Sorokin" formula.

--Dunbar

*In a subsequent issue of Blackjack Forum we tried to set the record straight. 10 years later I guess I'm still trying to set the record straight!

···

--- In vpFREE@yahoogroups.com, Dan Paymar <Dan@...> wrote:

At 9:41 PM +0000 11/4/09, nightoftheiguana2000 wrote:
>It's all about the size of your bankroll. Which is why the Kelly
>system is so important. The Kelly system produces the maximum
>average bankroll growth. The rich get richer, the poor get poorer,
>and the middle steadily disappears.

The Kelly system can work fine for games where you can make
relatively small changes in your bets, allowing you to bet a
calculated percentage of your bankroll, but that is rarely possible
in video poker. The only situation I can think of would be a 100-play
machine that lets you play any number of hands.

Also, Kelly is not very good for games with widely skewed
probabilities and payoffs, including video poker.

The only reliable calculation for risk is the Sorokin formula, first
applied to video poker by (I think) jazbo, and included as a feature
in Optimum Video Poker.

Dan

--
Dan Paymar
Author of best selling book, "Video Poker - Optimum Play"
Developer of VP analysis/trainer software "Optimum Video Poker"
Visit my web site at www.OptimumPlay.com

"Chance favors the prepared mind." -- Louis Pasteur

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

Thanks for the correction and background. This "risk of ruin" calculation assumes you will play till your last penny is exhausted, which according to the "Kelly System" is folly once you cross the line of overbetting the "Kelly Bet" (which is approximately equal to edge/variance times your current bankroll), unless you're deemed "too big to fail" and can count on a government bailout. Under the "Kelly System", the risk of losing your entire bankroll is zero, even if you can't reduce your bet, you would simply no longer bet. Also, just as for the Jazbo/Sorokin "risk of ruin" formula, it is similarly possible to exactly calculate the Kelly number, for those so inclined. In this article, Jazbo provides a link to a Perl script to do so:

http://www.jazbo.com/videopoker/kelly.html

···

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

The risk formula which you and NOTI are refering to as the "Sorokin formula" would be better named the "jazbo/Sorokin formula". Since I am at least partially responsible for the mis-naming of the formula, I want to jump in here to try to set it straight.

The formula itself has been around for a long time. But jazbo was the first to recognize that it could be applied to video poker: jazbo sent a note to his listserv group about it in December of 1998.

A couple of months later, a Russian mathematics student, Evgeny Sorokin, posted a cryptic RoR question about a 3-payoff game on the old bjmath.com site. He answered the question himself a short time later. His posts languished for several weeks until I realized what he meant, and more importantly, realized that his solution could be applied to video poker.

I posted my "Eureka" moment on bjmath, and MathBoy quickly realized that cashback could be included. We wrote an article titled "Risk Of Ruin for Video Poker and Other Skewed-Up Games" that was published in the Fall 1999 (Vol XIX, #3) issue of "Blackjack Forum".

Neither MathBoy nor I had an inkling of jazbo's earlier contribution until after the article was written.* Nor did any of the 3-4 people to whom we sent a draft. That's understandable--we were all blackjack players who had only recently begun to even think about video poker.

MathBoy and I did not refer to the formula as the "Sorokin formula" in our article. We ackowledged Sorokin's contribution, but referred to the formula as the "generalized risk equation". However, I think it's appropriate to call its application to video poker the "jazbo/Sorokin" formula.

--Dunbar

*In a subsequent issue of Blackjack Forum we tried to set the record straight. 10 years later I guess I'm still trying to set the record straight!