Guess who answered my question without asking many data?
The main problem with sequencing aces is productiivity. How many sequences per shoe will you remember? Also, will you be playing head up, and are you able to spread from one or two hands to 4 or more without suspicion? In my experience, we got only about one ace per hour of play even under decent conditions.
Aces may be cut off from their key cards. (Depending on the shuffle this may or may not be a significant problem.) Also, if you do not remember the suits precisely or mix up the order of the keycards, you will make bets that have no ace afterwards. Also, how consistently and smoothly do the dealers shuffle the cards?
Also, how often will you mess up your hands in order to position the ace? Each bet you sacrifice is small, but if there are many of them it makes a difference.
Another issue is whether you bet when the last keycard has come out but no more. In this case, with a perfect shuffle, the ace is most likely to appear as the fourth card out, which would be the dealer's upcard.
Each ace you track is one that is unlikely to appear within a given round. So after the first round or two, you should subtract one from the count for each tracked ace, if the cards you've seen indicate no aces are imminent.
Here's a simple approach to computing your RoR. It neglects many things, like whether you are betting with the count or getting any kind of rebates, or refinements on how you bet depending on what keycards may yet show or how many cards past the last keycard you've seen. But most important of all, it assumes you never make a mistake.
I will assume that when you bet for the ace, you catch it 70% of the time, 10% of the time it goes to the dealer, and 20% of the time it "disappears".
I will assume that when you bet for the ace, you catch it 70% of the time, 10% of the time it goes to the dealer, and 20% of the time it "disappears".
Your EV for the round is .7*(.50*5000-.002(10000))+.2(-.002*15000)+.1(-.37*15000)= 1175, with an approximate standard deviation of 5000 * sqrt[.7(1.44+1.33+1.33+6*.5) + .3 (1.33*3+6*.5)] = 13300.
Your greatest standard deviation will be 150K at about 128 trials, so your RoR at that point would be 250/150 standard deviations out, or about 5%.
I would double this figure (at least), probably triple or quadruple it if you take a reasonable error rate into account.
.jpg)
