Confused about Kelly

I thought that betting 'Kelly' was betting that proportion of your bankroll that corresponds w/ your advantage over the house as indicated by the true count (roughly speaking). I do this, but limit my max bet to about 2% of my bankroll regardless of the TC.

However ... I read something a while back (BJMath, article by MathProf circa 1997 titled 'When and Why Kelly "Breaks Down"' that indicates that Kelly breaks down when the advantage gets 'large' (whatever large is).

Normally it would seem a moot point to me because of my 2% rule, but I was in a situation where I had to split 3's (4 times, and two of them were double down situations.) The TC was very high at the time (+6 in a 6-deck game), so I had my max bet out. But because of the splits and double downs I had 6 times my max bet out. (Oh, and I learned the hard way what size wager I'm able to emotionally tolerate ......). The dealer busted but I about crapped my pants that hand.

My question is what to do in a situation like this (excluding the emotional tolerance thing)? Should I continue to split and double down?

What you are really doing is using Kelly as some sort of risk of ruin equation, which is not what it is. It is a growth maximizing formula. Bet at a level you are comfortable with. If you are uncomfortable with the amount on the table following splits, resplits, and doubles then you should bet smaller, not decline the opportunities. You can use risk averse indeces that will call for a stronger edge when putting out more money, but you will still sometimes face these situations. Any time you consider not playing your hand properly it indicates that you are overbetting your psychological limit and should consider betting smaller.

Thanks for your advice. I'm not really intending us use Kelly as a RoR tool, but rather to maximize profit. However, I am combining other measures (namely, the 2% rule) as ceiling amounts to help mitigate
the odd situations where Kely indicates a large % of my BR. I'm not too comfortable putting 5% of my BR on the table (Maybe if my advantage over the casino was like 75%, but how often does that happen :) ). I usually bet TC/2 - 0.5 to determine my bet amount, but I vary it up or down a bit to make things look more natural.

I also do odd things with the bets as part of cover. I'm confident I'm fooling the pit, but my concern is about fooling the surveillance folk. Are there any places, links, articles that you can point me to with regard to cover bets/actions that specifically address defeating the surveillance dept?

How are you defining your bankroll? It almost appears as if you are talking about a session bankroll rather than a total playing bankroll. It would help if you would clarify.

"... I read something a while back (BJMath, article by MathProf circa 1997 titled 'When and Why Kelly "Breaks Down"' that indicates that Kelly breaks down when the advantage gets 'large' (whatever large is)."

I just read an old post (think it was by Steve Jacobs) discussing high-end Kelly breakdown in Blackjack, and I *think* he said it starts to break down around 10%. A simple example (one MP uses in his paper) is when you wager 25% of your bankroll...you will not always be able to fully resplit. And when you do not adhere to proper play there will be a price to pay, which will be proportionate to the size of your wager (a large chunk, in this case).

"Normally it would seem a moot point to me because of my 2% rule, but I was in a situation where I had to split 3's (4 times, and two of them were double down situations.) The TC was very high at the time (+6 in a 6-deck game), so I had my max bet out. But because of the splits and double downs I had 6 times my max bet out."

Hey, it coulda been 8.5 max bets ;-). Seriously, this situation has nothing to do with anything in MP's paper. With a max bet capped at 2% of your BR you don't ever have to worry about topping out, no matter how many re-splits. Of course, you don't always carry you're whole BR around with you in your back pocket, so a good compromise is to keep a couple (extra) big bets handy for these fairly rare situations.

One more thing, I liked this comment by DD' so much I thought it was worth repeating...

"Any time you consider not playing your hand properly it indicates that you are overbetting your psychological limit and should consider betting smaller."

Excellent advice.

ANS

No, I'm talking total BR, which is all my cash in a seperate account used solely for gambling (Blackjack only). I don't have any real rules on session BR; some people only take a % of their total BR to the tables. Although I don't literally stuff my pockets with all my BR, I have access to it all if I wanted. I usually play until I get distracted (tired, etc) or feel heat. sometimes a session is 45 minutes, other times a session is 2 hours. Doesn't really matter to me session-wise; I'm concerned about the long run, like, where will my BR be in 5 years, 10 years, etc.

My gambling BR is totally excess cash. I could burn it, use it for booze, give it away, whatever, and not have a 100% loss impact me in a tangible way. I could lose it 3x's and not be affected. Don't get me wrong, I'd be seriously bummed out but I'd still be paying my mortgage and bills, saving for retirement, taking vacations. I live a very very very average lifestyle and far below my means.

BTW, I appreciate the input. I'm making money (after logging several hundred hours this year), but I don't consider myself a real bad-ass counter yet. I'm learning new stuff all the time, and as I get better I need to know more about camo (Green chip now, but BR allows higher when I'm more skilled.)

Thanks, advice well taken. I played that hand correctly (and other correctly were I lost several max bets in a row), but I was trying to be honest about the crapping my pants thing. Maybe I'm odd but I seem to be able to 1) freak out emotionally but 2) bet/play the right way. I also don't do things like chase losses, etc. I actually think this helps my cover a bit because I honestly look like a ploppy when this happens. I actually got up and did the happy dance after the dealer busted. The pit people were rolling on the floor laughing at me. They made fun of me for quite some time.

There is a difference between still being able to play correctly, and maintaing your emotional health. I think the advice to stay within your psychological bet limit regardless of BR is good.

When you do lose 6 max bets on one hand like that, it could take you many, many hours to get it back. That is the tough thing for me emotionally.

(after logging several hundred hours this year)

It's only mid-February. Did you mean in the last 12 months?

I have not read the entire thread so others may have said the same thing that I'm about to. If you don't want to bet more than 2% of your bank on an initial wager then you should not bet Kelly, but rather a fraction of Kelly, perhaps half Kelly. Putting this cap at a point where your utility calls for a bigger bet creates a non-optimal spread which creates a longer long run index and does not grow the bank as efficienctly. I'd advise looking at the Kelly optimal bet at the TC of +5 for the games you commonly play. This is the TC of greatest profitability in terms of action * ev * frequency. See how this corresponds to your 2% max bet ROR formula. Use a fraction of Kelly for all your bets that corresponds to the same ratio. Example, ev/var calls for a fraction of your bank that comes to $300, but you limit yourself to bets of $200, then try betting 2/3 Kelly for all your bets down the scale. This will create an optimal spread.

The only time when this is not a good plan is when there is a very large bankroll, ruin is no real concern, and the player is constrained by a low max bet. For example, he'd like to frequentl bet $1000 but the table max is $200. This would not call for reducing all his bets by 80%, but he'd just jump to the max sooner. He'd have a higher ev but a worse DI and longer long run index.

Thank you for your input. If I understand correctly, doesn't a TC of +5 roughly correspond to a 2% advantage for the player? If so, that falls nicely with my 2% cap. Perhaps I need to look at the % of RoR when betting > 2% of my BR and determine if the increased risk is worth the potential return. Normally my bet spread goes as follows:

TC | % of BR
<= -2 ..... leave table*
<= 1 ...... min bet
2 ......... .5%
3 ......... 1%
4 ......... 1.5%
>=5 ....... 2%

I've seen shoes with the TC >5, but they're fairly rare, and spreading more than 12 to 1 will probably bring heat. I would think at some point I'd have to constrain my max bet so as not to get kicked out. I also vary my betting amounts a bit so as not to look like a robot.

* I may sit out a hand or make a excuse to skip a hand and watch the count if its early in the shoe. I'll rejoin if the TC rises to acceptable levels. I wong a bit when I can.

I've only been seriously doing this for a few months, and appreciate your input.

Thanks again,

JS

I believe you should still use smaller fractions, perhaps .75 Kelly. the difference in growth is minimal and you will have less volatility and a smoother ride. It may avoid your having the difficult situations with lots of money on the table. Those are the very situations you should be excited about as you have the chance to make lots of money. That's when you need to get the money pushed out as fast as possible before they change dealers and break the deck, or decide to change the cards, or any of the other things that can mess things up. The optimal fraction for non-instantaneous resizing is also smaller than for continuous instant resizing. Resizing when you win or lose 20% of the bank, for example, would call for slightly smaller fractions which would increase the growth rate of your bankroll.