Probability Odds - Simulations

I made a program on SmallBasic which would simulate hands using the following criterias

Player Doubles 11 vs all, 10 vs 9 or lower, 9 vs 5 or 6
Player Doubles S19 v 6, S18 v 6 and lower, S17 v 3-6, S15&16 v 4-6, S13&14 v 5-6
Player Splits aces and eights always and nines vs all but 7 A 10, sixes versus 6 or lower, fours vs 5 or 6 and twos and threes versus 6 and lower
Player always Hits S17 if double is not available
Player Hits S18 v 9 or higher
Player stays with 12 or less vs. 6 or less unless its 12 vs 2 or 3.
Player Hits 16 or lower vs 7 or higher and always stays on H17.

Only one split aloud and no double after split
Only one card when you split aces
No Surrender

With the program you can view below I entered in 128 faces and 32 of each other card (or 8 deck shoe)
Always reshuffling after each hand and 8 million simulations I came across the following:
-A 0.11% player edge with the following stats:
(numbers in millions)
REGULAR REC 2.661-3.118-0.534 -0.52664
DOUBLING REC 0.396-0.229-0.047 +0.335026
SPLITTING REC 0.202-0.180-0.025 +0.021313
BLACKJACK REC 0.361-0.362 -.017 +0.179562
RETURN +9261.50 (+0.1158%)


Maybe I made a mistake, and I'll keep an eye on code and any errors I may have, but I wonder what Decision/Rules they use to say its a -0.65% play at this spot or wonder why my numbers a little off.

Has anyone else ran numbers similar to mine before as well?

CODE ON SMALLBASIC for my program I made
GFP855

You have mistakes in your splitting strategy and on your doubling on 9.

The mistakes I am looking for are math calculating mistakes that may give inaccurate results.

I have not simulated that situation since 2002 and forget what the play is with an A8 double vs 6. I Know for sure it is marginal to stay and double and will run that play at a later date in simulation.

The situations I am taking are from the Google Image search of "Blackjack perfect strategy" and have outlined the one in different colors and says to stand on H16 vs. D10 if you have three cards and hit if you have it on two which makes no sense. Bottom line is if any play is a losing prop, then the results could only be better when changing the play.

When I simulated 16 vs 10 in 2002 I figured it was so close to the same loss on both plays that it becomes truly count dependent thus staying on rich decks and hitting on weak ones. Also doubling is stronger in rich decks when you have a 40%+ chance of getting a ten or ace.

In spots where you are -14 or worse on the count (-2 pure) the net result will end with scoring 20/21 less often and because most double ups occur when the dealer has the cards six and lower the odds they bust go way down. In rich decks these doubles can win 2/3 of the time, weak decks at around less than 3/5 and the first hand of shoe it's about 64% winners softly in the middle of the two.

It appears you are attempting to "refine" basic strategy. While a noble task, you should be aware that the answers to these questions have been asked and answered countless times and simulated "billions" (anything results siming less than 400 million is suspect).

My advice would be to become familiar with the Qfit.com suite of products and acquire and read Blackjack Attack 3rd Edition as it is a required reference material for such questions.

Quote: Stealth

My advice would be to become familiar with the Qfit.com suite of products and acquire and read Blackjack Attack 3rd Edition as it is a required reference material for such questions.

Another Qfit plug by a new member. What are the odds?

Pretty funny if you ask me

Sounds like a fun exercise. Tried writing a sim before but always pissed me off because I wanted it to be really efficient (run fast / clean) but also customizable (change whether stand or hit S17 for dealer....BJ payouts....etc.). Argh it's frustrating.


Can you do a thing to print each hand individually, and maybe look through to see if any hand doesn't make sense as to why player win/loss/push, as well as payout?

Is -0.11% you calculated based on initial bet or initial bet + splits & doubles?

Player automatically winning dealt 21's vs dealer non-21? Or is it allowing dealer to continue to hit until 17+? Does dealer auto win with dealt 21 vs player dealt non-21?

Perhaps keep a record of how frequently each card is being used....make sure each card is equally likely to be dealt. ie: May have messed up on the function where you take a card out of the shoe....may have made it so it can still be dealt again OR it could be taking out a card from not being able to be dealt that hadn't been dealt.

ie: pick a num from 0 to 311. Say it's #22. Card[22] gets dealt. If your code now picks a ran num from 0 to 310.....that 312'th card (Card[311]) doesn't have the opportunity to be dealt.

OP is simulating basic strategy with one of the conventional set of BJ rules and has not claimed that the program is changing HIT/Stand/Double/Split decisions as part of any count. Thus, no "error" in strategy, i.e., in player decisions, can result in a player advantage of 0.11%.

This is almost certainly a programming error. This calculation has been done many time by countless people. There is no harm in one more smart person engaging the problem and working it through - but there are three certainties in life: Death, taxes and the House Advantage in Blackjack if you don't count cards.

Obviously another exercise in Futility.

Quote: cwwbjr

Obviously another exercise in Futility.

I don't think so.

Quote: Wizard

Another Qfit plug by a new member. What are the odds?

Stealth is a long standing good member of BJTF and in my experience a straight up decent guy just trying to give advice. Not so sure about the other accounts, but I will vouch that he meant no harm by the plug.

Quote: Wizard

Another Qfit plug by a new member. What are the odds?

I responded to a post by someone trying to re-invent the wheel. If his objective is to learn how to re-invent a wheel, then I apologize. If his answer is to use a wheel to solve his problem, then I stand by my recommendation.

My comment was not intended to plug anything, it was to provide the OP a tool to use in his efforts.

If your sarcastic comment was directed at me, then suggest you PM me with specific concerns.