Kelly Has Left the Room!

[blackjack avenger]

I'm bored soooo:joker::whip:

When contemplating betting in the real world Kelly betting probably should not be considered because betting Kelly requires resizing your bet after every hand played. Also, table min and max bets interfere with proper bet sizing, one can't bet pennies.:joker::whip:

The above does not mean we can't use elements of Kelly wagering.

 

[SleightOfHand]

The above does not mean we can't use elements of Kelly wagering.

Precisely. Just because we can't use perfect Kelly betting, does not mean that we cant do as best as we can. Kelly is the optimal way to bet for maximum bankroll growth. Personally, I make a betting ramp of .6 Kelly (~3.5% RoR) prior to a session and resize after every session. My BR is sufficiently big as to not have to worry about table mins and too small to worry about table max.

 

[Kasi]

I'm bored soooo:joker::whip:

When contemplating betting in the real world Kelly betting probably should not be considered because betting Kelly requires resizing your bet after every hand played. Also, table min and max bets interfere with proper bet sizing, one can't bet pennies.:joker::whip:

The above does not mean we can't use elements of Kelly wagering.

OK BA lol.

Great question.

There's more, for starters, I don't understand about "Kelly" betting than I do understand lol .

I guess, like you say, the purest of pure "Kelly" beeting would mean being able to bet pennies on each and every hand, even if $7.62 happened to be between table min and max.

That one would never bet on a -EV hand.

That ROR would be 0%.

That betting this way would maybe maximize the logarhythmic growth of your roll but maybe not maximize one's expectation.

But then there's "fixed-spread Kelly" betting like maybe more for real-life BJ wherein one bets in such an optimal way that one also is maximizing growth to roll. Including betting in -EV situations. But now ROR is not 0% it's 13.5%.

Betting "optimally" always sounds like such a great idea, doesn't it? Sometimes, I'm not even sure what that means. Or how, or if, or why, it relates to "Kelly" in the first place. What makes an "optimal" spread "optimal"?

I've always thought a spread that minimizes doubling-roll time and minimizes N0 time (might be the same thing, not sure) is "optimal". Something that achieves the best ratio of EV to SD over all other possible, or practical, given being able to bet either $5 or $10 etc, bet spreads.

To me, like you say, maybe "Kelly" is just another of these "theoretical" concepts, almost like "ROR", maybe like N0, that aren't able to be exactly implemented in real life but, yet, may have some bearing on "real-life" nonetheless.

My suggestion is move this question to the boring-for-most-people "Theory and Math" section wherein I won't feel asking the lots of questions I have on the subject won't bore the card-counters here.

 

[SleightOfHand]

OK BA lol.

Great question.

There's more, for starters, I don't understand about "Kelly" betting than I do understand lol .

I guess, like you say, the purest of pure "Kelly" beeting would mean being able to bet pennies on each and every hand, even if $7.62 happened to be between table min and max.

That one would never bet on a -EV hand.

That ROR would be 0%.

That betting this way would maybe maximize the logarhythmic growth of your roll but maybe not maximize one's expectation.

But then there's "fixed-spread Kelly" betting like maybe more for real-life BJ wherein one bets in such an optimal way that one also is maximizing growth to roll. Including betting in -EV situations. But now ROR is not 0% it's 13.5%.

Betting "optimally" always sounds like such a great idea, doesn't it? Sometimes, I'm not even sure what that means. Or how, or if, or why, it relates to "Kelly" in the first place. What makes an "optimal" spread "optimal"?

I've always thought a spread that minimizes doubling-roll time and minimizes N0 time (might be the same thing, not sure) is "optimal". Something that achieves the best ratio of EV to SD over all other possible, or practical, given being able to bet either $5 or $10 etc, bet spreads.

To me, like you say, maybe "Kelly" is just another of these "theoretical" concepts, almost like "ROR", maybe like N0, that aren't able to be exactly implemented in real life but, yet, may have some bearing on "real-life" nonetheless.

My suggestion is move this question to the boring-for-most-people "Theory and Math" section wherein I won't feel asking the lots of questions I have on the subject won't bore the card-counters here.

I could be completely wrong, but I believe Kelly betting produces a 13.5% RoR if you do not change your bet ramp. As for "optimal" betting, that is a little more subjective as to what your desires are (reliability:N0, growth speed:Kelly/SCORE, win rate). Regardless, these theoretical concepts are the best way to find your desired "optimal" bet. While you cannot perform these bets perfectly, as they typically require constant resizing of bets, getting as close as possible will do better than any other option.

 

[Grisly Dreams]

"I could be completely wrong, but I believe Kelly betting produces a 13.5% RoR if you do not change your bet ramp."

I think that Kelly betting, properly construed, would require that one constantly alter one's unit size. Whatever risk of ruin exists, would seem to arise from inability to do actual Kelly betting.

 

[zengrifter]

I could be completely wrong, but I believe Kelly betting produces a 13.5% RoR if you do not change your bet ramp.

Full '1' Kelly no resize double or bust. zg

 

[blackjack avenger]

Semantics or Confusion?

I think there is a problem in terminology that confuses many including myself in the past, or maybe even now!:joker::whip:

To me as soon as you say the word "Kelly" you bring up the theory of optimal bets and constant resizing.

If you say 13.53% fixed ror, this is not Kelly. This is just a fixed ror just like 10% or 5% ror. There is nothing magical about a 13.53% ror.

I don't know how, but if the moderator wants to move this thread then that is fine by me.

 

[blackjack avenger]

Optimal is Opitmal

Betting "optimally" always sounds like such a great idea, doesn't it? Sometimes, I'm not even sure what that means. Or how, or if, or why, it relates to "Kelly" in the first place. What makes an "optimal" spread "optimal"?

What makes an optimal spread optimal?:joker::whip:
bank*advantage/variance=optimal bet

 

[Kasi]

If you say 13.53% fixed ror, this is not Kelly. This is just a fixed ror just like 10% or 5% ror. There is nothing magical about a 13.53% ror....

Well, there has to be something a little "magical" about it since "=exp(-2)" = 13.53%. It assumes the best rate of growth, I guess, just like "pure full-kelly does", I think maybe lol.

I'd call this "Kelly-equivalent risk of ruin” when someone chooses to never resize again.

I think it may represent that even a constantly-resizing "pure full-Kelly bettor" will always have that ROR from that point forward should he choose to never re-size again from that point forward.

The trouble with "full-Kelly", as you know, even if one does constantly re-size, are the huge swings in roll it creates with high liklihoods. Once one has lost a bunch, with constant re-sizing, it could take forever to win it all back but the growth rate to roll would still be maximized.

Maybe lol.

 

[Kasi]

What makes an optimal spread optimal?:joker::whip:
bank*advantage/variance=optimal bet

And what makes an optimal spread optimal if betting with a spread at all counts?

And why is not simply maximizing EV, sounds good, an optimal spread of another definition?

 

[blackjack avenger]

The Elements of Kelly Do Not Leave the Room

And what makes an optimal spread optimal if betting with a spread at all counts?

And why is not simply maximizing EV, sounds good, an optimal spread of another definition?

Sure, if you want to play all or have a defined 1 to 10 spread then one can have an optimal bet ramp given those constraints, but that is probably not Optimal.:joker::whip:

So one can have Optimal
or
optimal with restraints:joker::whip:

 

[blackjack avenger]

12, 13, 14

Well, there has to be something a little "magical" about it since "=exp(-2)" = 13.53%. It assumes the best rate of growth, I guess, just like "pure full-kelly does", I think maybe lol.

I disagree, what is the difference between 12%, 13,53% and 14% ror, they each represent the chance of losing all if you don't resize. Where is the magic?:joker::whip:

Now optimal theoretical Kelly there is magic:
If you bet less your money will grow less rapidly given same ror, theoretical 0%.
If you bet .5 or 1.5 kelly you will have the same growth but wilder swings with 1.5 kelly.
If you bet 2 Kelly you will have no bankroll growth
If you bet over 2 kelly your bankroll will shrink

 

[blackjack avenger]

Agree Errrrrrr Ummmmmmm

I'd call this "Kelly-equivalent risk of ruin” when someone chooses to never resize again.

No major problem but I think the word Kelly should be used carefully.

One could say 13.53% fixed:joker::whip:

 

[Kasi]

I disagree, what is the difference between 12%, 13,53% and 14% ror, they each represent the chance of losing all if you don't resize. Where is the magic?:joker::whip:

Now optimal theoretical Kelly there is magic:
If you bet less your money will grow less rapidly given same ror, theoretical 0%.
If you bet .5 or 1.5 kelly you will have the same growth but wilder swings with 1.5 kelly.
If you bet 2 Kelly you will have no bankroll growth
If you bet over 2 kelly your bankroll will shrink

Well, lol, I guess the magic is only the 13.53% guy is maximizing growth of roll. The 12% guy is sacrificing growth rate for less volatility. I guess they are betting fractions of "kelly"?

While maybe a constantly re-sizing pure Kelly bettor may have an ultimate risk of losing everything of 0%, maybe, for him, a risk of "unhappiness" can be explored. Along with how long it might take him to dig out of his hole. Not that I have a clue about that lol. But, wouldn't the pure Kelly guy have a 50% chance of losing half his roll? And by then wouldn't it take him pretty long to get back to 100% of his roll? I think you'd agree anyway, one main drawback of "full-pure-kelly" is the wild volatility?

If you could pure Kelly bet, say any amount with pennies, would you choose to that or do the fixed-bet thing with 10 or 20K?

Just BS'ing lol. I'll never really understand it lol.

 

[Kasi]

\One could say 13.53% fixed:joker::whip:

Nah - you have to have "kelly" in there somewhere so at least one knows it's all based on the best way to bet your money taking risk and reward into account, don't you think? :grin:

 

[blackjack avenger]

Well, lol, I guess the magic is only the 13.53% guy is maximizing growth of roll. The 12% guy is sacrificing growth rate for less volatility. I guess they are betting fractions of "kelly"?

Which style maximizes log growth?
13.53% fixed ror
or
kelly resizing
I don't think they both can:joker::whip:

While maybe a constantly re-sizing pure Kelly bettor may have an ultimate risk of losing everything of 0%, maybe, for him, a risk of "unhappiness" can be explored. Along with how long it might take him to dig out of his hole. Not that I have a clue about that lol. But, wouldn't the pure Kelly guy have a 50% chance of losing half his roll? And by then wouldn't it take him pretty long to get back to 100% of his roll? I think you'd agree anyway, one main drawback of "full-pure-kelly" is the wild volatility?

If you could pure Kelly bet, say any amount with pennies, would you choose to that or do the fixed-bet thing with 10 or 20K?

Just BS'ing lol. I'll never really understand it lol.

Yes, in an alternate reality with no limitations:
lifetime
betting
table min and max
ability to calculate actual advantge and bet

I would bet optimally. I do have to live up to my name:joker::whip: