Bob Dancer's LVA - 19 MAY 2015
Going Broke
http://www.lasvegasadvisor.com/bob_dancer/2015/0519.cfm
*************************************************
This link is posted for informational purposes
and doesn't constitute an endorsement or approval
of the linked article's content by vpFREE. Any
discussion of the article must be done in
accordance with vpFREE's rules and policies.
*************************************************
I use FPDW as a common example, I leave it up to others to translate to their particular situations. Even today it is still a game that can be found, so it is not irrelevant to video poker. True, the earnings rate at quarters is on par with the minimum wage for a skilled player, but it might beat driving a bus from the airport to casinos as you mentioned as a backup plan. On the plus side the alcohol is free to FPDW players and there is no blood test or fbi background check required and you get to set your own hours.
Bob wrote: "Most of these formulae assume that we have a fixed bankroll --- with no extraneous inputs (perhaps income from a job, alimony, Social Security, a pension, etc.) or outputs (rent, our car needs a new transmission, expensive hobbies, etc.)"
Other inputs and outputs can be modeled as changes to EV. This is frequently put forth as a weakness to the Kelly system, but it's not at all. Of course there are other inputs and outputs, and they do count, model them as changes to EV. If your gambling is putting $10/hour into your bankroll and you put another $10/hour into your bankroll from your day job, your net EV into your bankroll is $20/hour.
Bob wrote: "I play in a world of murkier numbers."
Actually, we all do. Even the FPDW player is unlikely to get the computer optimal EV, instead there is some error rate which results in an actual EV of something less than computer optimal. The random number generator in the machine is not a true random number generator, it is a pseudo-random number generator, close but again some murkiness is involved. This is an issue with using the Kelly system, the common workaround is to bet less than Kelly so there is some margin of error. Typically applying math to the real world requires using some margin of error.
Bob wrote: "I'm not aware of video poker players going broke because they were expertly playing too big a game --- and got eaten alive by the variance. The ones I've heard of going broke are the ones who are no longer playing with an advantage. "
There's that word "advantage" again. I assume you meant "playing with an edge". I know a number of gamblers who went broke by gambling over their head (another way to say gambling over their bankroll) even though they clearly had an edge. It's surprisingly easy to do, in my experience. On thing video poker and poker has is variance, variance in spades you could say. I'm not saying you can ignore edge, edge is important, but so is variance. Assuming an edge, it's the variance that's the problem, it's the variance that can destroy bankrolls.
[Non-text portions of this message have been removed]
Bob wrote on going broke: "How much attention do I pay to the Kelly criterion when I play video poker today? NONE AT ALL." Sorry to hear that. Of course it's quite possible that your bankroll has become too large for video poker. You might need something that offers larger bet sizes. You could go into the stock market like Bill Gross. Bob wrote: "The numbers NOTI publishes concerning playing 25 cent FPDW forever are of no interest to me as a gambler (although they are of interest to me as a video poker writer and teacher). I am likely never going to play that game because 25 cent single-line games are of no interest to me whatsoever."
I should comment on the issue of accounting for all the other stuff you get for gambling. I recommend cash basis accounting, which means if something can get monetized and that money goes into your gambling bankroll, then it counts. Otherwise, if something can't get monetized it doesn't have any cash value in cash basis accounting. Most if not all gambling that I know of is done with cash, so cash basis accounting is probably the best fit. To give an example, if you were going to spend $10 from your day job for dinner, but instead won a free buffet coupon from the kiosk and put the $10 in your gambling bankroll, that counts, plus $10 to your bankroll. On the other hand if you won an unlimited pass to the steakhouse and ate so much you felt you should tip $20, $10 from your day job and $10 from your gambling bankroll, that counts also, but minus $10 from your gambling bankroll, so it ended up negative EV to your gambling bankroll.
[Non-text portions of this message have been removed]
Sorry to hear that. Of course it's quite possible that your bankroll has
become too large for video poker. You might need something that offers
larger bet sizes. You could go into the stock market like Bill Gross.
Why do you believe I need to make growing a bankroll a top priority in my life?
My bankroll is "large enough." (I hope)
My philosophy is to play with an edge and take Bonnie square dancing. I'm 68 years old, financially fine, with no heirs. I have my causes I'll donate to when I die, but whether I give these organizations $xxx or five times $xxx isn't that important to me. I don't envy the people who sit home and count their money every day.
My bankroll is not too large to play video poker. But it's large enough that the stakes I play for don't threaten my lifestyle.
My philosophy about such matters was different when I was 34 than when I'm 68. I made some good choices and saved well along te way, and I'm now reaping the rewards of that. And I like it that way!
Bob
[Non-text portions of this message have been removed]
hBob wrote on going broke: "How much attention do I pay to the Kelly criterion when I play video poker today? NONE AT ALL."
Bob wrote: "Why do you believe I need to make growing a bankroll a top priority in my life?"
You don't at all. Kelly's system only applies if you want to avoid bankruptcy at all costs while maximizing log bankroll growth. If these are non-issues for you, the Kelly system is a poor fit. In particular, if you are willing to deal with some risk of bankruptcy, the Kelly system is too conservative. And if your bankroll is infinite you can even play negative expectation games with no risk assuming a finite lifespan.
[Non-text portions of this message have been removed]
What if his bankroll is finite and lifespan is infinite?
---In vpFREE@yahoogroups.com, <nightoftheiguana2000@...> wrote :
You don't at all. Kelly's system only applies if you want to avoid bankruptcy at all costs while maximizing log bankroll growth. If these are non-issues for you, the Kelly system is a poor fit. In particular, if you are willing to deal with some risk of bankruptcy, the Kelly system is too conservative. And if your bankroll is infinite you can even play negative expectation games with no risk assuming a finite lifespan.
[Non-text portions of this message have been removed]
Well it was an interesting discussion the first few times. It would still be interesting if BD seemed to really want to be a part of this instead of getting dragged into it.
[Non-text portions of this message have been removed]
bobbartop wrote: "What if his bankroll is finite and lifespan is infinite?"
If bankruptcy is unacceptable: Kelly formula.
If some risk of bankruptcy is ok: ROR formula.
[Non-text portions of this message have been removed]
What "discussion"?
It appears to me that Bob is essentially asserting that, for the regular plays in which he engages, his ROR is near 0 and therefore bankroll considerations aren't a practical concern.
I'm willing to accept this at face value. For those who aren't, that doesn't mean that this is fodder for discussion. Both parties have to hold an interest in the topic for that to be the case. I gather Bob doesn't (his prerogative; and something I don't find unreasonable).
---In vpFREE@yahoogroups.com, <cdgnpc@...> wrote :
Well it was an interesting discussion the first few times. It would still be interesting if BD seemed to really want to be a part of this instead of getting dragged into it.
[Non-text portions of this message have been removed]
[Non-text portions of this message have been removed]
Is infinite bankroll with an infinite lifetime called Zombie video poker? I figured next you're gonna ask about finite bankroll with a finite lifetime, or video poker on the human scale.
For ROR over a particular timespan, I recommend Dunbar's Risk Analyzer.
For Kelly, Kelly actually goes one step at a time, or one hand at a time in the case of video poker, just like the 12 step programs, or Joe Walsh's song if you prefer that image, so Kelly is for humans, not zombies. Use the equation for approximate Kelly CE = EV - Variance/Bankroll/2 . EV, Variance, and Bankroll are in units of bets. Say, for example, you want to play one hand of $5 9-6 double-double at max bet ($25) and your current bankroll is $1000. You know you can't play computer perfect, but you're pretty sure you can play well enough to get the average casino hold down to 1.5% . So, Kelly approximate CE is -0.015 - 42/($1000/$25)/2 = -0.015 - 0.525 = -0.54 = -54%. -54% of $25 is -$13.50 . That's the approximate Kelly average value of playing this one hand, this -$13.50 has to come from somewhere, it either comes from your bankroll or it comes from your day job, or some other source if you have any. Obviously, if you want a fair bet for your bankroll size, assuming you put all the winnings into your bankroll, you would only bet if your CE was zero, or in this case someone else paid you the $13.50 cost to play this hand. If you wanted to grow your bankroll, assuming you put all the winnings into your bankroll, you would only bet if your CE was positive or in this case someone else paid you more than $13.50 to play this hand. CE can only be positive if EV is positive but a positive EV can actually be a negative CE for bankrolls less than infinity. Be careful of your estimate of EV, you could be wrong, so put some work into getting a good estimate or provide some extra margin of safety in your guess. Just assuming your EV is the computer perfect EV value is probably a mistake in most cases. Good luck.
[Non-text portions of this message have been removed]
The question about finite bankroll with infinite lifespan was for those of us who think Bob Dancer is God.
---In vpFREE@yahoogroups.com, <nightoftheiguana2000@...> wrote :
Is infinite bankroll with an infinite lifetime called Zombie video poker? I figured next you're gonna ask about finite bankroll with a finite lifetime, or video poker on the human scale.
[Non-text portions of this message have been removed]
BBT wrote: The question about finite bankroll with infinite lifespan was for those of us who think Bob Dancer is God.
Yeah, right.
The number of those people on this site matches the number of thumbs you have on your right foot!
[Non-text portions of this message have been removed]
All I know is I'm pretty sure I recall more than one casino player referencing God and Dancer in the same statement ...
- H.
---In vpFREE@yahoogroups.com, <bobdancervp@...> wrote :
BBT wrote: The question about finite bankroll with infinite lifespan was for those of us who think Bob Dancer is God.
Yeah, right.
The number of those people on this site matches the number of thumbs you have on your right foot!
[Non-text portions of this message have been removed]
[Non-text portions of this message have been removed]
Just another observation, maybe useful to others:
Approximate Kelly value = EV - Variance/Bankroll/2.
For large bankroll, Kelly value approaches EV.
For small bankroll, Kelly value approaches -Variance/2.
The crossover is at Bankroll = Variance/EV/2 at which the value is 0 for positive EV games and double the EV for negative EV games.
So, for example, a 99% return game has an approximate Kelly cost of 2% at a bankroll of Variance/EV/2 bets.
[Non-text portions of this message have been removed]
This is probably way past beating a dead horse, but I would like to point out that the approximate Kelly cost formula works for breakeven games as well as positive and negative games:
For a breakeven game, the approximate Kelly cost is just Variance/Bankroll/2. So for example, say, once you adjust for play errors, you're still skilled enough to play 10/6 double double for breakeven against the casino:
For a 100 royal bankroll, approximate Kelly cost is 0.105%
For a 10 royal bankroll, approximate Kelly cost is 1.05%
For a 5 royal bankroll, approximate Kelly cost is 2.1%
That's assuming the casino won't kick you out for winning, if the casino does kick you out for winning you get to leave a winner, lucky you.
[Non-text portions of this message have been removed]