Counting versus bangers...for ETF, Cacarullo, Wattenberger, SBA fans etc.

I mentioned in a previous thread how Wong never went from using basis strategy as his base for his non-counting tactics. He never apparently made a study of counting versus warps.

I have been playing with hacking into various simulators, by changing the hold card dealing sequences in them, for various ways that banger play might predict with different probabilities and accuracies tens, aces and low cards. A remarkable simplification has emerged, SUCH that it is best not to detail my results (Oh Clarke, You screwed up your copy of SBA again. Shut up!---example). I will probably repost it later on bjmath.com

If you have ever noticed the pattern of indexes, when rounded in various ways, the line for each total, versus different hole cards, itself seems to also provide a clue for how an index for the weighted probability of the dealers total might change.

Strangely a simple formula seems to provide indexes for different probabilities of the hole card being a ten: propability of the prediction of the dealer having a ten (similar for being pat) multiplied by half the index for that total versus a 7 upcard. You add to the index, when you have a banger prediction, of a ten, half of the index of that total versus 7 ( for 7s boost the index by 50%) multiplied by the accuracy you have noted for your predictions (to restate it slightly differently).

Mark this a could be?????

"Strangely a simple formula seems to provide indexes for different probabilities of the hole card being a ten: propability of the prediction of the dealer having a ten (similar for being pat) multiplied by half the index for that total versus a 7 upcard. You add to the index, when you have a banger prediction, of a ten, half of the index of that total versus 7 ( for 7s boost the index by 50%) multiplied by the accuracy you have noted for your predictions (to restate it slightly differently)."

Huh? I'm kind of tired today and don't have the energy to do a translation into English. Sorry. Would you mind providing an example using actual card values and numbers, and I will see if I am mentally alert enough to follow along.

I'm positive there is useful information hidden there somewhere.

Uston +/- 14 v. 10 =6; 14 v. 7 =14 ~.85% ten prediction accuaracy, ten banger predicted 14 v. 10 ~18

The 7 is very important on 14 vs 10. Does this hold true for other hands?

14 is just an example. The pattern you see with index tables is not accidental. You have an expected value of the hole card and next card, and the actual weighted probabilities of those cards as well. What I am theorizing is that the XX total v. 7 indexes tend to give an adjustment number that works. Basic strategy presumes that tens are 16/52 probable---which unfortunately has been used improperly to imply that basic strategy is "just imagine the dealer's hole card is a ten." Nothing could be further from the truth. But when there is a warp or banger prediction, the conditional probabilty rises to whatever your prediction accuracy overall is.

It just appears that from some quick a dirty patching to various simulators, to redraw the hole card to match the prediction, that the index is about the same, as adding 1/2 of the product of the prediction accuracy and that totals v. 7 index, for what index you use not for a given total (AKA known hole card) of the dealer's first two cards, but one that has an artificial assigned probability of that hole card. I am using these tools to try to find an index that is not dealer's upcard + a ten, but dealer's upcard + a 85% probable ten. Only the authors of these simulators can really find out if this shortcut, which allows you to adjust your indexes and not have to carry really oodles of indexes in your head, really works!

Although it has been my experience that I either see the card or I don't. So it is either 0% or 100% accurate. However, with a tracked shuffle, I can see where this would be very useful on 'next' card play. Do I take the card or give it to the dealer?, etc.

Of course 16/52 misses. This is only for high predictions---something I noticed when trying to get weighted indexes and looking at the source indexes. Look the pattern of any decision tables does indicate how a set of indexes could contain information on how this count responds to other weightings.

I know you wouldn't try any hole card play with only ~35% accuaracy (~16/52). Yet you answered as if I were advocating that. The possibility I raise, and it seems to work with the simulators I have hacked, to have conditional dealer hole card probabilities, is that the combinatorial analysis that results in normal count indexes carries with it, information on how you change an index with any change in the probability in the hole card and next card. The way an index changes with the up-card next in the chart, is an approximation roughly of any change in the dealer's initial total too.

I just pointed out that the indexes for use against a dealer's 7, multiplied by 1/2 and the accuracy rate for ten predictions, seems to match the indexes you get directly when you change the probability of the dealer's hole card being a ten.

You did the usual: false trigger and dis routine.

because I was unsure about the patches to the simulators I made. You can simulate what index to use when you know what the dealer's hole card is but not what index to use with a second, other than the count, weighted probability of the hole card. The indexes are not the same; I only posted my initial observation that it seemed that with high prediction accuracy, in the sims I made, it could also be approximated in the way I posted on.

Glad you guys understood!!!!

you and others made criticizing GBV over his advantages given for knowing an ace was going to land on a given spot +/-6 cards. Of course aces are 1/13th of the pack, BUT KNOWING AN ACE IS GOING TO BE WITHIN A ~12 CARD SPAN IS NOT THE SAME, as the statement that on average there will be an ace within a ~12 card span. Even Don defended GBV on that point and I commend him highly for that (imagine that! Don was the open minded one there AND NOT amoung his clones--thanks again to Don and maybe he can help again with this point to you.).

So when you say that on average there is a 4/13ths probability of a hole card being a 10, THIS IS NOT THE SAME AS SAYING: my flash view/my banger bend/my warp read/ indicates that this ONE particular hole card is ~4/13 probably a 10.

The rest of your answer is ridiculous as well. You I know read, Winning Without Counting, way, way, way back when, and you have referenced Basic Blackjack several times. I know that you would never ever be making a hole card play on a prediction that had ~4/13 accuracy of a 10 card!!!

So re-read the original post and now I will expand on the potential shortcut with the total v. 7 upcard indexes. If you simulate a known hole card, ie knowing the dealers hole card, you have a stair-step pattern that is much the same as normal indexes. It is enough to postulate that all indexes contain information, on how you do the equivalent of moving to the right or left of your index chart, when you have some indication of the value of the hole card. My initial findings are that an approximation can be made from the 7 up card indexes. I would welcome the thoughtful input you and the simulation experts like yourself can provide!

Not demanding, just asking. A polite no even from simulation experts would help as well. Thankyou (assuming a little more thought this time in reply).

NO.

a) You have not presented any cogent theory on why this shortcut should work.
b) I can't imagine any sensible theory justifying it.
c) If this is a non-linear effect, why do you have a linear equation fitted to it?
d) If the theory doesn't work for P = 16/52, I have no faith it will work for all three of the following: P = 1, P = .85, and P = .7. What kind of linear function would behave that way?
e) If the theory only works for P = .85 you should have said so.
f) You just haven't provided enough evidence to make it worth my while to pursue this. If MathProf, or Peter Griffin asked me personally to check it out, I would try to make the time. No offense, but you're not in that category.

It's your theory -- YOU check it out.

ETF

This was an initial finding that might have been biased in the way I simulated a hole card with prediction errors.

Any hole card prediction with a statistically significant rated estimate of error would move an index. A shoe with a 4/13 shuffle point, and the prediction that just one ten will be before the cut card is enough to move an index in that the possiblity, slim as it is, that all the tens might be behind is part of any compostion prediction.

You also neglect how ANY hole card prediction in FACT has less variance by any known measures of variance, than ANY composition based prediction. Any prediction of a hole card value that is in any way more accurate than the compostion based one, will have a better SCORE or DI type measure, of any sort you care to examine.

The use of the 7 upcard indexes as an estimator is based on my own output. I don't care to name the simulators I have modified for this due to the way that some software authors have jumped in poluted threads like this, objecting to registration in pseudo-nyms. You can guess who I reffer to. Thus I put the findings out as I have.

Supose you were going to give advice to Igor, the new poster here dealing with how the VA messed up his eye surgery etc. For Igor seeing a 10 up is not seeing a 10, but a 90% probable 10, with other possibilities to the right of most charts. If you were giving him the best advice to still use indexes, on low vision days, you would advise him to play not a definite up card but a probable up card. The more uncertain, the more he would lean his indexes to the right of the chart.

Now much the same applies to prediction of the hole card in regard to indexes for when the dealer's hole card is known too---not double exposure but in a normal game. The difference is that we lean to the right of the chart when the hole card prediction of having a ten in the hole is MORE definite. But the same approximation appears to apply as we might advise Igor further. For the simulations I attempted, for a weighted prediction of a hole card with errors, a further approximation, appears to be to add half the product of the prediction accuracy (of tens in the hole etc.) and the 7 upcard index for your total.

That I hope explains this!!!

Instead of making plays based on hole card plays versus basic strategy. The simulator gurus seem to have other agendas here. I know the hole card plays are far more profitable, but only SOTSOG and alienated seem to get the point. My findings are that a simple approximation may work but that is only an initial finding and I don't want to get thread polluting flack over daring to register my software in a psuedo-nym, so I leave discussion of specific output to others. The post I made on how recomendations for sight impaired play can lead to other insights about playing an uncertain hole card should have brought more insightful responses.

It is about time this topic was taken up -- see the ridiculous debate on books and software over whether Grosjean or May is a better hole card player and the snipping from both sides trying to both attack and cozy up to Don. It is resolvable ONLY by reopening and precisely pinpointing the optimal strategy shifts, between COUNTING and hole card/location playing.

Now I see your e-mail address.