bj sims vs basic strategy

my bj sims is for the european blackjack

Six decks
Dealer stands on soft 17
Player may only double on hard totals of 9-11
Player may double after a split
no resplit
No-peek (player loses total bet, including after doubling or splitting, if dealer has blackjack)
early surrender (except with A)

at the same time 3 players follow their own strategy :

player 1 : play the standard basic strategy
player 2 : play the same except standing with 16 vs 10

(when surrender is not available)
player 3 : follow a Risk-Averse basic strategy

witch is near BS except:

- double 9 vs dealer 4-6 only

- double 10 & 11 vs 2-8 only

- never split 2,3,4

- split 6 vs 5,6

- split 8 vs 2-8

- split 9 vs 4-6

- stand with soft 18 vs 9,10 & A

- hit 13 vs 2

- surrender 8,8 vs 9 & 10

over 1000000 hands with a random shuffle the result is:

player 1: edge : -0.45 (the theorical edge)

player 2: : -0.39

player 3: : -0.63

How can we explain the result for player 2 where standing

with 16 (with more than 2 cards) vs 10 is a true BS

violation ?

player 3 reduce the variance for a give back of 0.2 to the casino,
a good deal for a non counter ?

thanks for your opinion.

Your sim has the player surrendering the two-card 16 v. 10, so when he stands on 16 v. 10, it's because the hand is comprised of 3 or more cards. And, in fact, composition-dependent basic strategy is to STAND on multi-card 16 v. dealer's 10.

There's no surprise here.

"player 3 reduces the variance for a give back of 0.2 to the casino,
a good deal for a non counter ?"

Depends on how much the variance is reduced! Take EV/SD (the DI, or Sharpe ratio) for each, and see which is better (higher).

Don

thanks for your reply Don,

>And, in fact, composition-dependent basic strategy is to STAND on >multi-card 16 v. dealer's 10.
>There's no surprise here.

It's a very big (bad) surprise for me,

I know that composition-dependent basic strategy allow
this, I use it for example when I play with CSM, but as far as I can
remember these deviations are allowed only directly after the shuffle
(according to 'the wizard of odds').

If not, all strategy available on the net and in some (most) BJ books
are incorrect or to be more precise incomplete.

When we know how important is the BS in our strategy ...!?

I think 16 v 10 is the only deviation in these playing condition ,right ?

>Take EV/SD (the DI, or Sharpe ratio) for each, and see which is >better (higher).

I have done this simulation with my own software,until now I havn't yet imlemented the SD but I have a good reason to do it now.

Please allow me to ask you another question:

by making some new simulations with this optimal play

I have the following result :

with 1 player

over 15 000 000 hands

EDGE : -0.49 %

DEALER BREAKING RATIO 29.6 %

with 2 players

over 9 000 000 hands

EDGE player 1 : -0.43 %

EDGE player 2 : -0.42 %

DEALER BREAKING RATIO 28.7 %

with 3 players

over 7 500 000 hands

EDGE player 1 : -0.43 %

EDGE player 2 : -0.44 %

EDGE player 3 : -0.43 %

DEALER BREAKING RATIO 28.3 %

with 4 players

over 6 200 000 hands

EDGE player 1 : -0.40 %

EDGE player 2 : -0.47 %

EDGE player 3 : -0.48 %

EDGE player 4 : -0.37 %

DEALER BREAKING RATIO 28.2 %

with 7 players

over 5 000 000 hands

EDGE player 1 : -0.38 %

EDGE player 2 : -0.37 %

EDGE player 3 : -0.40 %

EDGE player 4 : -0.35 %

EDGE player 5 : -0.29 %

EDGE player 6 : -0.38 %

EDGE player 7 : -0.35 %

DEALER BREAKING RATIO 28.2 %

When computing the DEALER BREAKING RATIO I don't take in
account all rounds where the dealer don't take second cards
(all players hands are gone)
For me the hand is unresolved (in a DBR point of view)
but some DBR > 28.2 !? strange !

I have also an EDGE who vary from
-0.29 (7 players) to -0.49 (1 player) and this for a game
with the same rules !?

Strange too isn't it ?

Thanks for all !

"I know that composition-dependent basic strategy allow this, I use it for example when I play with CSM, but as far as I can remember these deviations are allowed only directly after the shuffle (according to 'the wizard of odds')."

If you are a BS player (you're not counting) all rounds are identical for you. The expectation and play of the hand doesn't vary from one round to another, provided we do not consider the cut-card effect (we deal a fixed number of rounds per shoe). So, the recommnedation is valid off the top, or anywhere else.

As for your simulations, you don't state the standard error involved. There is no reason for the edges to vary or for the dealer break ratio to vary in the manner you describe. Try 500 million rounds for each sim and report back! :-) I suspect the differences will go away.

Don

> I suspect the differences will go away.

simulation done with more 50 000 000 hands

7 players:

p1: edge:-0.41

p2: edge:-0.42

p3: edge:-0.42

p4: edge:-0.45

p5: edge:-0.42

p6: edge:-0.44

p7: edge:-0.43

DEALER BREAKING RATIO : 28.2

with 1 player:

p1: edge:-0.45

DEALER BREAKING RATIO : 29.7

this confirm your point of view for the edge.

For the DBR,the number of hands played is not important,you reach very quickly these values (28.2 and 29.7).
As I have just some trouble with 1,2 and 3 players, j presume
the difference is due to hands where the dealer don't take
second cards(witch is exceptional with more than 3 players).
skipping the surrender option don't solve the problem.

With one player if I take in account these hands, j reach
a DBR of 22,8 (compare with 29,7 previously).

My conclusion is the DBR of 28.3 is only reached in situation where the dealer must take a second cards without exception.

your opinion ?

P.S. with more than 50 000 000 hands , standing with 16 (with more than 2 cards) vs 10 has shown no profit(same edge).

thanks.

a player who ALWAYS stands. this way the dealer will always have to take a card (unless this player who stands gets a blackjack!). might as well have two players who always stand and then see your results on the other players.

the cheeze

Nice idea toecheeze,

let' see

with one player I reach a dealer breaking ratio of ... 28.0 %

why 28.0 and not 28.2 (28.3) ?

because when the player has blackjack with a dealer card <> 10 or A
the hand is removed without giving him the possibility to stand.

with 2 players the DBR is OK (28.2).

The conclusion is clear(for me), the DBR of 28,3 is reached when
the dealer MUST take every time a second card,
if not, depending of the fact you take in account these hands,
you will find a DBR lower or greater (but never equal) than the one inspected.
The part between the inspected value and the one you reach will be function of the number of dealer hand 'unplayed'.

This is the kind of thinks you can learn from making your own simulation.

Pure convention, nothing else.

It's just like computing the edge where we only
take in account the initial money played and
never the one used for double and split...

pure convention...