thanks4thefish: Card composition playing strategy

[thanks4thefish]

I'm well aware this amounts to a meagre profit generally.......however.

On a single deck game, which is essentialy break even, deal 6/7 boxes 1 round only, a profit can only be generated with playing strategy deviation.

Whilst one could merely count, it would be more profitable if one could make a more informed decision based on the actual cards in play.

There's a few tips in Griffin's work eg. stand 16v10 if you see more 5's than 6's out, & counting would be fairly practical for 10's vs no 10's, but I'm looking for something more specific,
perhaps a broken down count for each major decision.

At first I thought it may be as simple as a balanced count of first draw cards that would bust you vs those that were most beneficial, but the count would also have to take into account the effects of removal on the dealers hand & to some extent further draws by the player & dealer, enter the computer.

Does anyone know of any work done in this area.
(I realise the fluctuations are huge, as I have simmed it with the Griffin count, but am hoping to both lift the TBA & lower the ROR thru a properly employed 'board composition playing strategy)

[ET Fan]

If you use a count with a high playing efficiency, it's already optimized for this.

Or are you looking for a computer to give you a playing efficiency of 100+%? That would be illegal on land, and might violate your TOA online.

ETF

> I'm well aware this amounts to a meagre
> profit generally.......however.

> On a single deck game, which is essentialy
> break even, deal 6/7 boxes 1 round only, a
> profit can only be generated with playing
> strategy deviation.

> Whilst one could merely count, it would be
> more profitable if one could make a more
> informed decision based on the actual cards
> in play.

> There's a few tips in Griffin's work eg.
> stand 16v10 if you see more 5's than 6's
> out, & counting would be fairly
> practical for 10's vs no 10's, but I'm
> looking for something more specific,
> perhaps a broken down count for each major
> decision.

> At first I thought it may be as simple as a
> balanced count of first draw cards that
> would bust you vs those that were most
> beneficial, but the count would also have to
> take into account the effects of removal on
> the dealers hand & to some extent
> further draws by the player & dealer,
> enter the computer.

> Does anyone know of any work done in this
> area.
> (I realise the fluctuations are huge, as I
> have simmed it with the Griffin count, but
> am hoping to both lift the TBA & lower
> the ROR thru a properly employed 'board
> composition playing strategy)

[Saboteur]

Check out www.beejack.com. In the lower left-hand corner of their page, download a demo version of their BJ analyzer. The demo doesn't include all possible hands; I assume the full "purchased" version does.

I have no idea what methodology they use to calculate the playing decisions, or even if they're correct.

Their site has message boards. You might find more details there.

[thanks4thefish]

"If you use a count with a high playing
efficiency, it's already optimized for this.

It would be optimised for all decisions combined yes, not for individual decisions, eg 16 v 10 with the Griffin count would have you counting both 5 & 6 as positive 1 despite 5 being the card you least want to see on the board & 6 the card you most want to see.

Similarly with 14v10 if no 8's or 9's were out on a full table (uncounted cards in most systems) it would probably be wiser to stand.

It might sound ridiculous to have a seperate count for each common important play, but remember with one round only you won't be keeping a running count anyway.

'Or are you looking for a computer to give
you a playing efficiency of 100+%? That
would be illegal on land, and might violate
your TOA online.'

Neither, just looking for some guidelines in how to factor in cards seen to playing strategy.
One of the reasons counting is so practical is that most mere mortals would be unable to remember how many of each card had come out.
Single deck one round only, you don't need to remember as all the cards out are in front of you...........just how to use the info?

[ET Fan]

Discussed in his book "Sklansky Talks Blackjack." I don't know how far he develops it there. Ultimately, what you could do is develop specialized counts for different decisions, based on the tables in Griffin. There's a little about it here: http://www.twoplustwo.com/sklanskybj.html

I think John Imming was one of the pioneers of this idea. You have to consider the effect of this on your speed of play, and your error rate.

ETF

> "If you use a count with a high playing
> efficiency, it's already optimized for this.

> It would be optimised for all decisions
> combined yes, not for individual decisions,
> eg 16 v 10 with the Griffin count would have
> you counting both 5 & 6 as positive 1
> despite 5 being the card you least want to
> see on the board & 6 the card you most
> want to see.

> Similarly with 14v10 if no 8's or 9's were
> out on a full table (uncounted cards in most
> systems) it would probably be wiser to
> stand.

> It might sound ridiculous to have a seperate
> count for each common important play, but
> remember with one round only you won't be
> keeping a running count anyway.

> 'Or are you looking for a computer to give
> you a playing efficiency of 100+%? That
> would be illegal on land, and might violate
> your TOA online.'

> Neither, just looking for some guidelines in
> how to factor in cards seen to playing
> strategy.
> One of the reasons counting is so practical
> is that most mere mortals would be unable to
> remember how many of each card had come out.
> Single deck one round only, you don't need
> to remember as all the cards out are in
> front of you...........just how to use the
> info?

[thanks4thefish]

"Discussed in his book "Sklansky Talks
Blackjack. I don't know how far he
develops it there."

Thanks, I'll keep it in mind. I had heard that book was relatively worthless, & even plain wrong in many parts. Most books only briefly touch on card composition strategy & then only based on one hand. the terminology of 'key card' in relation to composition strategy does sound hopeful however. I guess another book can't hurt, I've got most of 'em anyway LOL.

"Ultimately, what you could do is develop specialized counts for different decisions, based on the tables in Griffin."

That's my goal yes. I'm thinking that the effects of removal would imply different values to the same card for different decisions though, so don't know how useful the tables would be.
12v4 the 8 & 9 would be desirable, but not for 14,15,16v 10 for example.

I had thought about simming with a modified deck, but I know errors can occur due to not also removing a subset of random cards. Random requiring many sims with different subsets I assume & then taking the average result............starting to get a little complex, for me anyway

"There's a little about it here:
http://www.twoplustwo.com/sklanskybj.html

Thanks I'll have a look

"I think John Imming was one of the pioneers
of this idea. You have to consider the
effect of this on your speed of play, and
your error rate."

Thanks ETF, if in doubt I'd revert back to a decision based on the Griffin count. As for speed in play, I'd be interested having obtained the info. to perhaps break it down to simple rules for each decision based on key cards present or absent.

I guess I was just hoping someone had already done this painful process & published.

[thanks4thefish]

"Check out www.beejack.com. In the lower
left-hand corner of their page, download a
demo version of their BJ analyzer."
>Their site has message boards. You might
>find more details there.

I've got SBA which allows me to sim for card composition strategy based on the hand in question but it doesn't factor in the other cards on the felt. I'll check it out though, thanks, if nothing else you've given me a new BJ site to peruse.

[ET Fan]

I hadn't heard what you said about Sklansky being "plain wrong." It was offered by Snyder in BJF, so couldn't be too bad. He gives intuitive "reasons" for basic strategy plays, which might not be 100% technically correct, but they aren't meant to be.

I don't know what you mean by "the Griffin count." What I am talking about is many counts. Ie. one for every strategy decision. I'm talking about using the strategy tables in Chapter 6 of ToB to derive such counts. I don't know what else you could be referring to.

There are a few things you can do to adjust hi-lo, such as adjusting for 6s as negative instead of positive for hard 16 v ten, and aces as positive for insurance. MathProf wrote about some of these on bj21 green. Going much farther beyond that naturally becomes very complicated, which is why nobody has written about it.

ETF

[Brick Waller]

7,8,9(core cards) side count using hi-opt with a side count of aces would be very beneficial for what you're trying to achieve,but very difficult to master.

[thanks4thefish]

"I don't know what you mean by "the
Griffin count."

4,5,6,7 +1, 10's -1, (chapter 4, optimal systems for variation of strategy)

"What I am talking about is many counts.
Ie. one for every strategy decision."

Me too!, I'm thinking a simple balanced 1 level count based on rounding the most extreme card values, (most beneficial vs most harmful.)

"I'm talking about using the strategy tables in Chapter 6 of ToB to derive such counts."

Just had a look & you are 100% correct. I had only glanced the 16v10 & 14v4,9 mini-table in Chapter 3

Can't get more specific than chapter 6 tables!, I can make the count as simple or as complex as I want by factoring only the most important cards or all of them & pivoting at 0 for decision making, definately will go for the former

If only using a few cards I could use a multi level count rather than rounding too extremely to 1 level, or having to use fractions.

Great book, despite the fact I can only understand a fraction of it! Prior to owning a sim I used to use the chart in chapter 11 to work out basic strategy for differing no. of decks.

I guess I'm guilty of skipping over a lot of the tables in BJ books & focussing on the text, assuming they're too complex, but it's actually fairly straight forward. Rather than trying to look at the tables as a whole I've gotta read the intro (a coupla times) & then look at a few decisons that interest me, that I know the rough answer for already so I can tell if I'm misinterpreting.

Might even dig Don's latest work out this week & give the table a try instead of glossing over them.

Thanks

[Saboteur]

You're welcome!

The software gives you the "correct" play based on deck-composition, not just hand composition. It was probably created in response to the advent of online blackjack, the idea being that it helps you keep track of "known" cards all the way through to the next shuffle, not just for the current hand.

Your response leads me to wonder if you'd noticed that. I may have misunderstood you, though.