Online blackjack and index play
In those games, even if the deck is shuffled after every hand, I'm wondering if it's possible to sit at 3rd base, evaluate the cards seen by players before you, and then make index plays to shift the edge toward you. In other words, off the top of the deck, what's the EV difference between a 1st base BS player and 3rd base index player (I18 or full)? I don't have time to code up a sim and I haven't been able to find that data online anywhere.
Or maybe I have -- in the Wizard's Blackjack Calculator, it lists two figures: Optimal and Realistic results. Do the Optimal Results correspond to a player at 3rd base at a full table playing full indexes, vs. the "Realistic Results" which is a basic strategy player?
Note: I'm assuming that online games still can't be beaten with index play even if it's possible to shave off a few tenths from the edge because otherwise there would be bots all over the place doing just that. But what's the EV difference?
I would assume not only multiple decks and shuffle after each hand, but individual shoes for each player.
Quote: MathExtremistI can't bet online due to my state of residence, but I had a thought this morning. Most online casinos offer pretty liberal blackjack games because they're computerized and can easily shuffle between hands (which they do). Several of these also offer multi-player games with timers similar to online poker -- you wait for the person before you to act, then it's your turn, then it's the next player's turn, etc.
In those games, even if the deck is shuffled after every hand, I'm wondering if it's possible to sit at 3rd base, evaluate the cards seen by players before you, and then make index plays to shift the edge toward you. In other words, off the top of the deck, what's the EV difference between a 1st base BS player and 3rd base index player (I18 or full)? I don't have time to code up a sim and I haven't been able to find that data online anywhere.
Or maybe I have -- in the Wizard's Blackjack Calculator, it lists two figures: Optimal and Realistic results. Do the Optimal Results correspond to a player at 3rd base at a full table playing full indexes, vs. the "Realistic Results" which is a basic strategy player?
Note: I'm assuming that online games still can't be beaten with index play even if it's possible to shave off a few tenths from the edge because otherwise there would be bots all over the place doing just that. But what's the EV difference?
This is called "depth charging" and has been described in a lot of the classic blackjack books. I believe I first read about it in Blackbelt in Blackjack by Snyder.
With liberal rules and a single deck, I believe you can gain the edge. There was a story in the Big Book of Blackjack (also by Snyder) of a guy pulling off a speed addict act who would walk into a casino with a paper shopping bag full of benjamins, play all 7 spots on a single-deck table, and bet increasing amounts on each spot ($100 on first base, $700 on third base). The story took place decades ago when single decks had better rules; I doubt you would be able to get away with this now. But if you knew all the single deck index plays, I'm sure you could theoretically amass a tidy +EV situation this way.
Anyway, for online play, I feel like it would be tough to get good enough rules and the right conditions (multiple players) to make it work. Also you would have to make sure each player doesn't have their own shoe, as the previous poster said.
Sorry, I'm at work and don't have my library handy, so I'm not sure what the theoretical edge is that was published.
and this passage about a single-deck game:
If I set up a seven-player table, with two rounds between shuffles, and I allow the third-base player, who is using the Zen Count, to see all of the cards of the players who play their hands before him, with a flat bet he will win at the rate of 0.77%. (This is with Reno rules, which allow doubling on 10 and 11 only, and the dealer hits soft 17; with downtown Las Vegas rules, where you can double on any two cards, but the dealer still hits soft 17, the flat-betting third base player will win at the rate of 1.06%, assuming he can see all of the other players' cards.)
With only one round between shuffles, you can only flat bet, so the .77% or 1.06%, depending on the rules, gives you an upper bound on the EV.
Quote: dwheatleyA quick search found this link
and this passage about a single-deck game:
If I set up a seven-player table, with two rounds between shuffles, and I allow the third-base player, who is using the Zen Count, to see all of the cards of the players who play their hands before him, with a flat bet he will win at the rate of 0.77%. (This is with Reno rules, which allow doubling on 10 and 11 only, and the dealer hits soft 17; with downtown Las Vegas rules, where you can double on any two cards, but the dealer still hits soft 17, the flat-betting third base player will win at the rate of 1.06%, assuming he can see all of the other players' cards.)
With only one round between shuffles, you can only flat bet, so the .77% or 1.06%, depending on the rules, gives you an upper bound on the EV.
Thanks. That's really interesting, though I'm not sure it's accurate. In a single-deck pitch game, cards come out face-down so the 3rd base player does not see "all of the cards of the players" -- just some of them (e.g. busts, splits, blackjacks, but *not* stands). I'd like to hear from the Wizard how he derived his "Optimal Results" statistic on the blackjack calculator. That has a much lower figure for 1D, DA2, H17 than +1.06%, though it may not be taking into account the expected visible cards from players 1-6 in the full-table, 3rd base scenario. Is that the discrepancy?
Two questions for follow-on study:
1) What is the edge of a player using perfect composition-dependent strategy in a 1D pitch game at 3rd base/full-table who sees a normal number of other player cards but not "all of the cards of the players"? I'd guess it's a lot lower than the cited statistics.
2) Assume a single-deck online casino. Is there a way for the 3rd-base player to implement a pseudo-count (using a computer) which takes into account (a) the prior player's cards actually seen, and (b) the likely distribution of cards not seen, based on the information that was learned from the prior players' actions? For example, two random cards have an unknown count. But if the player at 1st base stands vs. dealer 9, it is unlikely that the count on those cards is +2. More likely it is -1 or -2. Now, you'd have to be pretty quick on your feet to apply Bayes like this in a live casino setting, but it's trivial to write software for it. Has this concept been studied?
Quote: dwheatleyA quick search found this link
and this passage about a single-deck game:
If I set up a seven-player table, with two rounds between shuffles, and I allow the third-base player, who is using the Zen Count, to see all of the cards of the players who play their hands before him, with a flat bet he will win at the rate of 0.77%. (This is with Reno rules, which allow doubling on 10 and 11 only, and the dealer hits soft 17; with downtown Las Vegas rules, where you can double on any two cards, but the dealer still hits soft 17, the flat-betting third base player will win at the rate of 1.06%, assuming he can see all of the other players' cards.)
With only one round between shuffles, you can only flat bet, so the .77% or 1.06%, depending on the rules, gives you an upper bound on the EV.
Wow, that passage was lifted verbatim out of Blackbelt in Blackjack. I'm surprised Snyder or his goons haven't brought that site down for copyright infringement.
Quote: MathExtremistThanks. That's really interesting, though I'm not sure it's accurate. In a single-deck pitch game, cards come out face-down so the 3rd base player does not see "all of the cards of the players" -- just some of them (e.g. busts, splits, blackjacks, but *not* stands).
The link addressed this, saying the analysis was only accurate if 3rd base could somehow see all the cards. It would be tough to pull off in the real world (easy with 6 confederates :)), let alone online.
Quote:I'd like to hear from the Wizard how he derived his "Optimal Results" statistic on the blackjack calculator. That has a much lower figure for 1D, DA2, H17 than +1.06%, though it may not be taking into account the expected visible cards from players 1-6 in the full-table, 3rd base scenario. Is that the discrepancy?
If I understand the difference between the 2 numbers from the Wiz correctly, it doesn't have anything to do with counting cards, but means using the much more complicated charts like this that take into account the exact composition of your hand, and not just the total.
And his circle of 13 distort probailities. A winning edge in 13 cards is a definite loser with 52 cards. And that is the minimum number in any casino, brick and mortar or on-line.
