New basic strategy

Hey Guys,

I just published a new basic strategy for french casinos (dealer stand on soft 17). It's on github, I can't publish link, repository is black-jack-resolver /juliendubost/black-jack-resolver)

I didn't considered the possibility to double after a split.
I'm surprised with results since it differs from the wizard basic strategy (/games/blackjack/strategy/4-decks/).
I use an infinite number of decks because I think it is unimportant on strategy if you play with at least 4 decks, which is totally standard in France (more than 6 decks).

For example if you look at the pocket 8, it is slightly different, same for a soft 18.

I computed the total game EV:
=> 1.057 using wizard strategy
=> 1.062 using mine

Definitely a difference that can matter on the long run.

I was wondering if the methodology used by the wizard is known ? So that i could understand from where this difference come from.

Thanks a lot.

If the Wizard's methodology is anything like mine (note that I never surrender), it's something like this:

First, calculate the probability that the dealer will end up with 17, 18, 19, 20, 21, or bust based on the dealer's hole card, but exclude 2-card 21s.

Next, calculate the player plays.
Start with hard 21; compare the EV of standing against hitting - in this case, since the EV of hitting is -1 and the EV of standing > -1, the play is Stand.
Calculate the value for hard 20; if you hit, 1/13 of the time you will have hard 21 and 12/13 of the time, you will bust, so compare the sum of the EVs of those two possibilities, each of which is multiplied by the probability of hit happening, to the EV of standing.
Repeat for hard 19, hard 18, and so on down to hard 4. (A hard 3 is impossible.)
Next, calculate soft 21; hitting is 1/13 x the EV of hard 12 + 1/13 x the EV of hard 13 + ... + 1/13 x the EV of hard 20 + 4/13 x the EV of hard 21; compare this to the EV of standing on 21.
Calculate soft 20: hitting is 1/13 x the EV of soft 21 + 1/13 x the EV of hard 12 + ... + 4/13 x the EV of hard 20.
Repeat through soft 12.
Now, calculate the value for doubling on hard 21 (remembering that your bet is now double), and if it is better than the hit/stand EV, the strategy for hard 21 becomes Double.
Repeat for hard 20, 19, and so on through hard 4.
Do the same for soft 21 through soft 12.
Finally, to determine whether or not to split Pairs, take a pair of 8s as an example.
Calculate 1/13 x the EV of soft 19 (i.e. you have 8-A) + 1/13 x the EV of soft 20 + ... + 1/13 x the EV of hard 17 + 4/13 x the EV of hard 18, then multiply the sum by 2, and if it is higher than the highest of the hit / stand / double values for (in this case) hard 16, the strategy becomes Split.
If you want to include Surrender, I assume that you can Surrender on any hand where the EV < -0.5.

Quote: AllanKicker

I was wondering if the methodology used by the wizard is known ? So that i could understand from where this difference come from.

Thanks a lot.
link to original post

Possibly helpful:
https://youtu.be/jCF-Btu5ZCk

Here is the link to OP's basic strategy.

OP's link to his basic strategy

This "new" basic strategy has some differences on whether to double 10, and 11 vs a dealer Ten and Ace. I suspect it is because AllanKicker's new strategy looked at doubling without taking into account that dealer has peeked at his facedown card and determined that he does not have BJ -that extra piece of info make doubling vs T, A much more attractive.

The OP states on github:
“Total expected value using this strategy is: 1.062074
(you win a total of 1.062074 every time you do an initial bet of 1)”
That makes bj a losing game for the casino -or does it? Simply playing basic strategy? Sorry I have serious doubts. The game would have ceased to exist soon after it was introduced.

Maybe the correct formulation would be “you LOSE a total of 1.062074 every time you do an initial bet of 1”. In that case, Wizards basic strategy performs better: “you LOSE a total of 1.057 every time you do an initial bet of 1”.

The presentation of the BJ basic strategy is not easy to learn. It has too many entries. It should be consolidated. Instead of learning a multitude of individual parameters, they should be grouped based on a common decision. For example, all 2-card hands ‘less than or equal to 8’ require a common decision: Hit (in the Stand/Hit table). Or, all 2-card hands ‘greater than or equal to 12’ require ‘Do not double down’ in the corresponding chart.

There are such tables on the internet. I understand links are not allowed here.

Quote: gordonm888

Here is the link to OP's basic strategy.

OP's link to his basic strategy

This "new" basic strategy has some differences on whether to double 10, and 11 vs a dealer Ten and Ace. I suspect it is because AllanKicker's new strategy looked at doubling without taking into account that dealer has peeked at his facedown card and determined that he does not have BJ -that extra piece of info make doubling vs T, A much more attractive.
link to original post

Thanks for the link (it does not point to the root repository but that's ok ! )

Yes, I do not assume dealer has peeked at his face-down because in France this does not exists, dealer have only 1 face up card before serving players.

I now understand better why the wizard strategy is more likely to double VS T or A, this make sense.

Consequently, my result of 1.057 EV for wizard strategy is wrong, I do not know how much it is wrong, it is probably near the reality but yes I under estimate value.

Quote: toolyp

The OP states on github:
“Total expected value using this strategy is: 1.062074
(you win a total of 1.062074 every time you do an initial bet of 1)”
That makes bj a losing game for the casino -or does it? Simply playing basic strategy? Sorry I have serious doubts. The game would have ceased to exist soon after it was introduced.

Maybe the correct formulation would be “you LOSE a total of 1.062074 every time you do an initial bet of 1”. In that case, Wizards basic strategy performs better: “you LOSE a total of 1.057 every time you do an initial bet of 1”.

The presentation of the BJ basic strategy is not easy to learn. It has too many entries. It should be consolidated. Instead of learning a multitude of individual parameters, they should be grouped based on a common decision. For example, all 2-card hands ‘less than or equal to 8’ require a common decision: Hit (in the Stand/Hit table). Or, all 2-card hands ‘greater than or equal to 12’ require ‘Do not double down’ in the corresponding chart.

There are such tables on the internet. I understand links are not allowed here.
link to original post

From my computation I can confirm the game is EV+ using basic strategy, meaning that yes, it is theoretically a LOOSING game for the casino.

I don't know in your country but in mine, in France, most of the players do not respect the basic strategy, a LOT of players are standing at 16 against a 7, 8, 9, T or A on the bank, etc ...

So, a possibility is that, theoretically it's a loosing game for casino but in practice, it's a wining game due to players game style.

Most links^* are allowed once new member probation is over.

https://wizardofodds.com/games/blackjack/basics/#wizards-simple-strategy

Quote: Dieter

Most links^* are allowed once new member probation is over.

/games/blackjack/basics/#wizards-simple-strategy

*

There are some rules, like no free advertising.

link to original post

No offence but I saw better presentations of BS on 3 separate charts. Color red indicates the corresponding action: split, double, hit. Very easy to memorize.
Is reddit ok to post links>
Blackjack subreddit

Quote: toolyp

Quote: Dieter

Most links^* are allowed once new member probation is over.

/games/blackjack/basics/#wizards-simple-strategy

*

There are some rules, like no free advertising.

link to original post

No offence but I saw better presentations of BS on 3 separate charts. Color red indicates the corresponding action: split, double, hit. Very easy to memorize.
Is reddit ok to post links>
Blackjack subreddit
link to original post

I have my own disagreements with that chart, so I made my own. (For example, I code split as V, like the hand signal.)

I don't think the link posted correctly.

Quote: Dieter

Quote: toolyp

Quote: Dieter

Most links^* are allowed once new member probation is over.

/games/blackjack/basics/#wizards-simple-strategy

*

There are some rules, like no free advertising.

link to original post

No offence but I saw better presentations of BS on 3 separate charts. Color red indicates the corresponding action: split, double, hit. Very easy to memorize.
Is reddit ok to post links>
Blackjack subreddit
link to original post

I have my own disagreements with that chart, so I made my own. (For example, I code split as V, like the hand signal.)

I don't think the link posted correctly.
link to original post

The fewer entries the easier. Maybe this link works –
https://www.reddit.com/r/blackjack/comments/hbo4qo/any_advice_on_memorizing_basic_strategy_on_black/

Quote: Dieter

Quote: AllanKicker

I was wondering if the methodology used by the wizard is known ? So that i could understand from where this difference come from.

Thanks a lot.
link to original post

Possibly helpful:

link to original post

Ok, so now I know, it comes from this excel sheet.
Thanks a lot for this.

Quote: ThatDonGuy

If the Wizard's methodology is anything like mine (note that I never surrender), it's something like this:

First, calculate the probability that the dealer will end up with 17, 18, 19, 20, 21, or bust based on the dealer's hole card, but exclude 2-card 21s.

Next, calculate the player plays.
Start with hard 21; compare the EV of standing against hitting - in this case, since the EV of hitting is -1 and the EV of standing > -1, the play is Stand.
Calculate the value for hard 20; if you hit, 1/13 of the time you will have hard 21 and 12/13 of the time, you will bust, so compare the sum of the EVs of those two possibilities, each of which is multiplied by the probability of hit happening, to the EV of standing.
Repeat for hard 19, hard 18, and so on down to hard 4. (A hard 3 is impossible.)
Next, calculate soft 21; hitting is 1/13 x the EV of hard 12 + 1/13 x the EV of hard 13 + ... + 1/13 x the EV of hard 20 + 4/13 x the EV of hard 21; compare this to the EV of standing on 21.
Calculate soft 20: hitting is 1/13 x the EV of soft 21 + 1/13 x the EV of hard 12 + ... + 4/13 x the EV of hard 20.
Repeat through soft 12.
Now, calculate the value for doubling on hard 21 (remembering that your bet is now double), and if it is better than the hit/stand EV, the strategy for hard 21 becomes Double.
Repeat for hard 20, 19, and so on through hard 4.
Do the same for soft 21 through soft 12.
Finally, to determine whether or not to split Pairs, take a pair of 8s as an example.
Calculate 1/13 x the EV of soft 19 (i.e. you have 8-A) + 1/13 x the EV of soft 20 + ... + 1/13 x the EV of hard 17 + 4/13 x the EV of hard 18, then multiply the sum by 2, and if it is higher than the highest of the hit / stand / double values for (in this case) hard 16, the strategy becomes Split.
If you want to include Surrender, I assume that you can Surrender on any hand where the EV < -0.5.
link to original post

Yes it is exactly what I did, I implemented it as fully deterministic so, probability of hitting any non-ten-valued card is 1/13 and hitting a ten-valued card is 4/13 at any time.

Then you know the bank's final scores probabilities given the face-up card, and so on.

To add robustness and avoid mistakes, I implemented monte-carlo tests where I simulate a lot of cards draw to validate all the deterministic values of stand EV and bank's final scores probabilities.

Quote: AllanKicker

Quote: toolyp

The OP states on github:
“Total expected value using this strategy is: 1.062074
(you win a total of 1.062074 every time you do an initial bet of 1)”
That makes bj a losing game for the casino -or does it? Simply playing basic strategy? Sorry I have serious doubts. The game would have ceased to exist soon after it was introduced.

Maybe the correct formulation would be “you LOSE a total of 1.062074 every time you do an initial bet of 1”. In that case, Wizards basic strategy performs better: “you LOSE a total of 1.057 every time you do an initial bet of 1”.

The presentation of the BJ basic strategy is not easy to learn. It has too many entries. It should be consolidated. Instead of learning a multitude of individual parameters, they should be grouped based on a common decision. For example, all 2-card hands ‘less than or equal to 8’ require a common decision: Hit (in the Stand/Hit table). Or, all 2-card hands ‘greater than or equal to 12’ require ‘Do not double down’ in the corresponding chart.

There are such tables on the internet. I understand links are not allowed here.
link to original post

From my computation I can confirm the game is EV+ using basic strategy, meaning that yes, it is theoretically a LOOSING game for the casino.

I don't know in your country but in mine, in France, most of the players do not respect the basic strategy, a LOT of players are standing at 16 against a 7, 8, 9, T or A on the bank, etc ...

So, a possibility is that, theoretically it's a loosing game for casino but in practice, it's a wining game due to players game style.
link to original post

In America the vast majority of BJ players know basic strategy very well. It’s striking to me that evn the dealers help those who dont know basic strategy. Some dealers say “The kook recommends to do this, not what you did.” Other players say “She’s right”.

Of course, the card counters alter BS sometimes. Thats how they expose themselves. They would split T+T v 5/6 or stand on 15 v 10.

I’m telling you, they would have rule out the game of blackjack. BS is very easy to learn and use and the casinos would have lost lots of money by now.

Quote: toolyp

In America the vast majority of BJ players know basic strategy very well. It’s striking to me that evn the dealers help those who dont know basic strategy. Some dealers say “The kook recommends to do this, not what you did.” Other players say “She’s right”.

Of course, the card counters alter BS sometimes. Thats how they expose themselves. They would split T+T v 5/6 or stand on 15 v 10.

I’m telling you, they would have rule out the game of blackjack. BS is very easy to learn and use and the casinos would have lost lots of money by now.
link to original post

This is interesting.
Maybe I miss something, I'll continue to work, especially on implementing the estimation of a double after a split EV.
And checking the impact of using X decks instead of infinite number of decks.

But this should be a minor impact on game EV.
I tested a lot and I'm pretty sure of what I computed.

Quote: AllanKicker

Quote: toolyp

The OP states on github:
“Total expected value using this strategy is: 1.062074
(you win a total of 1.062074 every time you do an initial bet of 1)”
That makes bj a losing game for the casino -or does it? Simply playing basic strategy? Sorry I have serious doubts. The game would have ceased to exist soon after it was introduced.

Maybe the correct formulation would be “you LOSE a total of 1.062074 every time you do an initial bet of 1”. In that case, Wizards basic strategy performs better: “you LOSE a total of 1.057 every time you do an initial bet of 1”.

The presentation of the BJ basic strategy is not easy to learn. It has too many entries. It should be consolidated. Instead of learning a multitude of individual parameters, they should be grouped based on a common decision. For example, all 2-card hands ‘less than or equal to 8’ require a common decision: Hit (in the Stand/Hit table). Or, all 2-card hands ‘greater than or equal to 12’ require ‘Do not double down’ in the corresponding chart.

There are such tables on the internet. I understand links are not allowed here.
link to original post

From my computation I can confirm the game is EV+ using basic strategy, meaning that yes, it is theoretically a LOOSING game for the casino.

I don't know in your country but in mine, in France, most of the players do not respect the basic strategy, a LOT of players are standing at 16 against a 7, 8, 9, T or A on the bank, etc ...

So, a possibility is that, theoretically it's a loosing game for casino but in practice, it's a wining game due to players game style.
link to original post

AllanKicker,

toolyp is correct: for the rule set you described, B.S. is NOT a winning strategy for the player. Did you forget to consider the effect of a dealer BJ when you calculated the EV?

Dog Hand

Quote: DogHand

AllanKicker,

toolyp is correct: for the rule set you described, B.S. is NOT a winning strategy for the player. Did you forget to consider the effect of a dealer BJ when you calculated the EV?

Dog Hand
link to original post

I'm currently looking at the wizard basic strategy sheet (link in the youtube video /watch?v=jCF-Btu5ZCk)

But from which source do you affirm that it is NOT a winning strategy for the player ? Is it the last page of the google sheet (see below) ?

Quote:

No dealer BJ 0.04024768481
Winning dealer BJ -0.04509646021
Grand total -0.004848775402


Dealer BJ 0.04733727811

I do not know how to interpret that lines, it's not covered in the video

I've only looked at the green table in the github link and it looked relatively correct except it must have had peeked (as has splitting AA vs A) but not sure why not D11vs10 and some soft doubles (c.f. https://wizardofodds.com/games/blackjack/strategy/8-decks/ )

One thing you may have overlooked is that because these tables assume Dealer has peeked you cannot just assume that your EV vs Ten or Ace is asis. You need to factor that a proportion of hands will have lost to Dealer BJ before you had a chance to act.

The other factor, I don't know French BJ rules, is whether you are allowed to Double after Split, this affects things like splitting 2s vs 2.

Quote: charliepatrick

I've only looked at the green table in the github link and it looked relatively correct except it must have had peeked (as has splitting AA vs A) but not sure why not D11vs10 and some soft doubles (c.f. https://wizardofodds.com/games/blackjack/strategy/8-decks/ )

One thing you may have overlooked is that because these tables assume Dealer has peeked you cannot just assume that your EV vs Ten or Ace is asis. You need to factor that a proportion of hands will have lost to Dealer BJ before you had a chance to act.

The other factor, I don't know French BJ rules, is whether you are allowed to Double after Split, this affects things like splitting 2s vs 2.
link to original post

Scroll up, most of this has been covered.

Quote: charliepatrick

I've only looked at the green table in the github link and it looked relatively correct except it must have had peeked (as has splitting AA vs A) but not sure why not D11vs10 and some soft doubles (c.f. /games/blackjack/strategy/8-decks/ )

One thing you may have overlooked is that because these tables assume Dealer has peeked you cannot just assume that your EV vs Ten or Ace is asis. You need to factor that a proportion of hands will have lost to Dealer BJ before you had a chance to act.

The other factor, I don't know French BJ rules, is whether you are allowed to Double after Split, this affects things like splitting 2s vs 2.
link to original post

The github link assume there is no peeke from the dealer (french rules).
French rules allow double after split but for the moment the github does not compute this.

The example you raise (D11 VS 10) is interesting and the answer is in the EV table for a 10 on the bank, here is what we learn from it:
- Hit then stand with 11 against ten on the bank has an expected value of 1.006 (= a net gain of 0.006 = total net gain of 0.012 after bet is doubled)
- Hit one or multiple cards then stand at the best position has an expected value of 1.033 (= net gain of 0.033).

Because 0.033 is greater than 0.012 => best move is to hit, not to double, despite the fact that double is still EV+.

I played a French blackjack game. It was a CSM using 6 decks of cards. It was Hit-17, DAS, no dealer peek for blackjack, and player losing all busted hands.

Hey, interesting results! The slight difference in EV is intriguing, especially over the long term. It makes sense that you didn't include doubling after a split, if that's not common in the French rules. The infinite deck assumption seems reasonable given the standard 4+ decks in France. As for the magician's methodology, there may be specific assumptions or rounding methods that could explain the variation. Perhaps an examination of his published work could provide some clue

Quote: cliper1

Hey, interesting results! The slight difference in EV is intriguing, especially over the long term. It makes sense that you didn't include doubling after a split, if that's not common in the French rules. The infinite deck assumption seems reasonable given the standard 4+ decks in France. As for the magician's methodology, there may be specific assumptions or rounding methods that could explain the variation. Perhaps an examination of his published work could provide some clue
link to original post

Bolding added by moderator.

Your reference to the Wizard as "the magician" is very telling; I strongly suspect you of being an AI bot.

Quote: gordonm888

Quote: cliper1

Hey, interesting results! The slight difference in EV is intriguing, especially over the long term. It makes sense that you didn't include doubling after a split, if that's not common in the French rules. The infinite deck assumption seems reasonable given the standard 4+ decks in France. As for the magician's methodology, there may be specific assumptions or rounding methods that could explain the variation. Perhaps an examination of his published work could provide some clue
link to original post

Bolding added by moderator.

Your reference to the Wizard as "the magician" is very telling; I strongly suspect you of being an AI bot.
link to original post

Sorry, it's just my communication style) You can send me a captcha if you want :p

Quote: gordonm888

Quote: cliper1

Hey, interesting results! The slight difference in EV is intriguing, especially over the long term. It makes sense that you didn't include doubling after a split, if that's not common in the French rules. The infinite deck assumption seems reasonable given the standard 4+ decks in France. As for the magician's methodology, there may be specific assumptions or rounding methods that could explain the variation. Perhaps an examination of his published work could provide some clue
link to original post

Bolding added by moderator.

Your reference to the Wizard as "the magician" is very telling; I strongly suspect you of being an AI bot.
link to original post

It is likely google translation. “wizard = magician”. Anyway, no matter how you spin this new basic strategy, it still is a losing proposition. No matter who the “shaman” is, this B.s. will lose .06 for every 1 bet.

Quote: toolyp

[
It is likely google translation. “wizard = magician”. Anyway, no matter how you spin this new basic strategy, it still is a losing proposition. No matter who the “shaman” is, this B.s. will lose .06 for every 1 bet.
link to original post

Do we have a source about this .06 loss ?
Is it from a book or something ?
I would like to know about the method used to compute this.
Meanwhile I maintain it's a wining game while it has not been proved wrong.

I think an explanation could be about variance, because nobody want to play a high variance game when EV is below a positive threshold.
For example, you can play a game that is EV+ like 20% but if there is high variance, this mean you can loose a lot and for a long time before being positive in terms of earning.

I just updated repository with dealer's peeked option.

Here is what I have computed in terms of expected values using different strategies.

dealer's peeked	hit on soft 17	This resolver strategy EV	Wizard of the odds strategy EV using no double after split
		1.062291	not available
X		1.069313	1.067432
	X	1.060393	not available
X	X	1.068152	1.067637

I've explained the approach of EV tables and it's open for contributions so do not hesitate to ask questions and/or suggest things.

Quote: AllanKicker

Quote: toolyp

[
It is likely google translation. “wizard = magician”. Anyway, no matter how you spin this new basic strategy, it still is a losing proposition. No matter who the “shaman” is, this B.s. will lose .06 for every 1 bet.
link to original post

Do we have a source about this .06 loss ?
Is it from a book or something ?
I would like to know about the method used to compute this.
Meanwhile I maintain it's a wining game while it has not been proved wrong.

I think an explanation could be about variance, because nobody want to play a high variance game when EV is below a positive threshold.
For example, you can play a game that is EV+ like 20% but if there is high variance, this mean you can loose a lot and for a long time before being positive in terms of earning.
link to original post

Chaman, Google translated “wizard” to “magicien” in French. Allow me to call you “chaman”. Making your Bj basic strategy a winning system would represent the act of a shaman. To that end, the casinos would stop the game of blackjack as it is played today.

Julian Braun, a programmer from IBM, refined Edward Thorp's BJ strategy in the 1960s. Julian Braun is unanimously considered the “father” of blackjack basic strategy (B.S.). All subsequent varieties differ very little from the original. We dont talk here about variations created by card counting (indices).

It’s been widely accepted that Braun’s B.S. made BJ a 50-50 game as closest as it can be. But still a losing game. Otherwise it cant be offered in the casino. It’s the law.

Your latest blackjack basic strategy makes the game a winning proposition for the player. That is, a losing game for the casinos. Every B.S. player nonchalantly wins $60 for every $1000 bet. One million bj players a year, $60 million a year in losses for any big casino!

Even if the casinos were run by idiots, the law would stop such a game.

Maybe you didnt have a good translation of Julian Braun’s book or program. Seems to me you put the cart in front of the horse. That is, you calculated EV from dealer’s perspective. Your B.S. is +EV for the dealer, the casino, therefore a losing proposition for the player.

Casino winning percentages much higher even than every calculation of the house edge. The winning percentage is over 10% at blackjack. And you are talking about +0.06% for the player! You see the reality in the real world: -10% for the player.

As for “variance”, it is a smokescreen. It is the ubiquitous excuse when a system fails. “Play billions of hands to validate a system!” Thats gross BS, not B.S.!

Methinks we are “lost in translation” here, to quote an acclaimed movie.

No offence, mon ami.

Quote: toolyp

Chaman, Google translated “wizard” to “magicien” in French. Allow me to call you “chaman”. Making your Bj basic strategy a winning system would represent the act of a shaman. To that end, the casinos would stop the game of blackjack as it is played today.

Julian Braun, a programmer from IBM, refined Edward Thorp's BJ strategy in the 1960s. Julian Braun is unanimously considered the “father” of blackjack basic strategy (B.S.). All subsequent varieties differ very little from the original. We dont talk here about variations created by card counting (indices).

It’s been widely accepted that Braun’s B.S. made BJ a 50-50 game as closest as it can be. But still a losing game. Otherwise it cant be offered in the casino. It’s the law.

Your latest blackjack basic strategy makes the game a winning proposition for the player. That is, a losing game for the casinos. Every B.S. player nonchalantly wins $60 for every $1000 bet. One million bj players a year, $60 million a year in losses for any big casino!

Even if the casinos were run by idiots, the law would stop such a game.

Maybe you didnt have a good translation of Julian Braun’s book or program. Seems to me you put the cart in front of the horse. That is, you calculated EV from dealer’s perspective. Your B.S. is +EV for the dealer, the casino, therefore a losing proposition for the player.

Casino winning percentages much higher even than every calculation of the house edge. The winning percentage is over 10% at blackjack. And you are talking about +0.06% for the player! You see the reality in the real world: -10% for the player.

As for “variance”, it is a smokescreen. It is the ubiquitous excuse when a system fails. “Play billions of hands to validate a system!” Thats gross BS, not B.S.!

Methinks we are “lost in translation” here, to quote an acclaimed movie.

No offence, mon ami.

Thanks for this sources, I'll have a look to it.
Then, I'm not convinced as well by this result of positive EV for player because like you say, and your estimation's approach is correct, casino would have troubles running this game.

This said, reality can sometimes be surprising and I trust data more than the "casino can't run a loosing game" statement or "it's widely accepted" statement.
For now, the partially validated data I have show this game is loosing for the bank.

I'll continue to dig, especially by doing a full montecarlo model of this game, which is not a big deal, and hopefully it would be an irrefutable proof.

There is a lot of things that could explains why it's not worth for players to exploit a 6% EV game and why it can be worth for a casino to exploit a theoretically EV- game.
Examples:
- As a player exploiting an EV+ game can be less worthy than working a fulltime job at the minimum wage, because of variance & because of the loosing streak you must be able to absorb with you own cash.
- As a casino, having a theoretically EV- table game can be a way to attract people that will spend money on EV+ things like slots.
- As a casino in France, you are forced by the law to have table games if you want to exploit slots machines, pretty much all french casinos are loosing on blackjack and even on roulette because earnings do not cover human resources expenses.

Etc.

Quote: AllanKicker

Quote: toolyp

Chaman, Google translated “wizard” to “magicien” in French. Allow me to call you “chaman”. Making your Bj basic strategy a winning system would represent the act of a shaman. To that end, the casinos would stop the game of blackjack as it is played today.

Julian Braun, a programmer from IBM, refined Edward Thorp's BJ strategy in the 1960s. Julian Braun is unanimously considered the “father” of blackjack basic strategy (B.S.). All subsequent varieties differ very little from the original. We dont talk here about variations created by card counting (indices).

It’s been widely accepted that Braun’s B.S. made BJ a 50-50 game as closest as it can be. But still a losing game. Otherwise it cant be offered in the casino. It’s the law.

Your latest blackjack basic strategy makes the game a winning proposition for the player. That is, a losing game for the casinos. Every B.S. player nonchalantly wins $60 for every $1000 bet. One million bj players a year, $60 million a year in losses for any big casino!

Even if the casinos were run by idiots, the law would stop such a game.

Maybe you didnt have a good translation of Julian Braun’s book or program. Seems to me you put the cart in front of the horse. That is, you calculated EV from dealer’s perspective. Your B.S. is +EV for the dealer, the casino, therefore a losing proposition for the player.

Casino winning percentages much higher even than every calculation of the house edge. The winning percentage is over 10% at blackjack. And you are talking about +0.06% for the player! You see the reality in the real world: -10% for the player.

As for “variance”, it is a smokescreen. It is the ubiquitous excuse when a system fails. “Play billions of hands to validate a system!” Thats gross BS, not B.S.!

Methinks we are “lost in translation” here, to quote an acclaimed movie.

No offence, mon ami.

Thanks for this sources, I'll have a look to it.
Then, I'm not convinced as well by this result of positive EV for player because like you say, and your estimation's approach is correct, casino would have troubles running this game.

This said, reality can sometimes be surprising and I trust data more than the "casino can't run a loosing game" statement or "it's widely accepted" statement.
For now, the partially validated data I have show this game is loosing for the bank.

I'll continue to dig, especially by doing a full montecarlo model of this game, which is not a big deal, and hopefully it would be an irrefutable proof.

There is a lot of things that could explains why it's not worth for players to exploit a 6% EV game and why it can be worth for a casino to exploit a theoretically EV- game.
Examples:
- As a player exploiting an EV+ game can be less worthy than working a fulltime job at the minimum wage, because of variance & because of the loosing streak you must be able to absorb with you own cash.
- As a casino, having a theoretically EV- table game can be a way to attract people that will spend money on EV+ things like slots.
- As a casino in France, you are forced by the law to have table games if you want to exploit slots machines, pretty much all french casinos are loosing on blackjack and even on roulette because earnings do not cover human resources expenses.

Etc.
link to original post

Oh, I was too conservative! I said 0.06%. It’s a typo. It is s whopping 6%. Based on your statement:
“Total expected value using this strategy is: 1.062074
(you win a total of 1.062074 every time you do an initial bet of 1)”.

That 6% EV+ is better than what any (financial) bank in the world can afford to offer. The French casinos would be the most attractive banks in the world. They offer 6% gain on a very short term. With a bankroll of $10,000, you play daily and make $600 in profit.

I thought of a minimal bankroll for an American gambler spending a 3-week vacation in France. The $10,000 bankroll balloons to over $10,000 at the end of the vacation time! WOWW!! Think about a bankroll of $100,000….

Think about it, even if you, Allan, are a casino rep (or a rep of the association of the French casinos). Cuz thats what I think about you. They, the French Casino Assoc, decided: “Hey lets advertise in America. They have gamblers with lots of money. And they are attracted to France as tourists. Unfortunately the Americans avoid French casinos because they believe our blackjack game has bad rules (no hole card, etc.) Lets convince them that we discovered a surefire winning strategy. It is called “Magique” or “Chaman Stratégie. It guarantees 6% in profits. It is as easy to learn as the American estrategie but it is the only system that assures a big gain.”

Sorry, you wont convince anybody in America with the “merits” of your new B.S. It is a huge stretch. You saw the reactions here of American BJ players. There is no such strategy possible. You saw the winning percentages in blackjack in American casinos: 10%. It means -10% for the players, not +6%. Huge difference!

I suspect the French casinos report a higher winning percentage for blackjack, at least 11% (or -11% for the players). That’s cuz the rules of French blackjack are worse than in America caused by that second card of the dealer. Maybe they are getting worse, as the water of Seine and the ceremonies at the Olympics this summer.

No offence, mon ami. Lets take everything with a smile. I’ll forget about your B.S. right away…..



I won indeed $1000 with “Magicians” BJ online app. But it’s a totally different beast, not just B.S. See my posts in the “Card Counting Quirks” in this forum.

Update: i finally found the good link to wizard BS: games/blackjack/strategy/calculator/

=> We are near the exact same BS (mine and wizard's one) which is a good point.

Now i have to understand where I made a mistake (if I made one), I'll work on it.
But FYI I'm not saying that MY strategy is EV+, I say ALL BS are EV+ for player, but I gonna check this and publish methodology so no worry and no need to debate longer on this.


For a general discussion purpose, about the 10000$ bankroll, keep in mind you can loose 5k straight the first session on a +6% EV game, that's why it's not so easy to exploit a +6% EV game.

Poker players knows that, because of the variance, stress, emotion, etc, you can end being a low wage earner when you do the ratio (gain - loss) / time whith an average hedge of 5%.


Another example, some french casinos started to abandon the "stud poker game", I duno if you have it in US, a dealer told me this game was EV- and they affirm that based on their loosing amount over time, not because of a mathematical calculation.

Then, pretty much all casinos followed and this game has been abandoned everywhere. But, it has been running for years.

I also know most european casinos are loosing on table games, due to human resources expenses to run it, I duno in US but right now the argument of "casino would have abandon the game if it was EV-" does not convince me, because a lot of casinos voluntarily run loosing games for some good reasons and legal reasons as well (in France you are forced to run table games in order to exploit slot machines).

Blackjack has a very low variance compared to slots or video poker. Some of these games have 30 to 50 times the variance of blackjack. I typically play slots with an edge of 0.5-2%. I am killing these games. Sure, I have losing months, but I have yet to have a single losing quarter playing online.

I don't believe for a second that the casinos can offer +EV BJ games and not get crushed by AP player community. The casino would be lucky not to lose money almost every day.

You were right I've spotted a dramatic issue in my starting hands probabilities, the pocket ten-valued card was highly over estimated, leading in an over estimation of the EV.
computed EV using BS is still winning but 0.2% for European rules (no peeked) and without double after split allowed, which is still is an interesting results VS wizard of odds since their house edge calculator always give a little advantage to the house.
Using US style (dealer's peeked) and no double after split allowed => 0.7% EV

Still need a full probabilistic model to verify this on a big number of hands

Thanks for your patience :).

There is a really great open source BJ evaluator already available on github courtesy of Eric Farmer: https://github.com/possibly-wrong/blackjack

This has been tested by a number of experts and does finite deck calculations for a variety of rules. I found the code easy to understand and modify.

Quote: Mental

There is a really great open source BJ evaluator already available on github courtesy of Eric Farmer: /possibly-wrong/blackjack

This has been tested by a number of experts and does finite deck calculations for a variety of rules. I found the code easy to understand and modify.
link to original post

Yes I've seen it, I also found this one on githuib from a computer science researcher : hhoppe/blackjack, it validate the wizardofodds approach, I'll try it to test the BS. Sounds like very fast an optimized.