Updated blackjack and card counting simulator

Sorry, I hadn't replied sooner. I set the default count to Hi-Lo. I haven't set any strategy deviations as defaults. I don't know if I should do that, I am thinking of leaving it up to the user to pick which ones they actually do.

Thanks,

Phil

Sorry, for some reason I thought your post was from a couple of months ago. I guess I did reply sooner. lol. Thanks.

I’ve tested this simulator and found it reliable.  To test this, I used your default Zen values to do the simulation 7 times each, with or without insurance.

When taking insurance at Counts At Or Above Zen=+5, I found these numbers

% hand won: 43.466%

% hand lost: 47.997%

% hand pushed: 8.537%

When taking insurance at Counts At Or Above Zen=+50, I found these numbers

% hand won: 43.479%

% hand lost: 48.040%

% hand pushed: 8.480%

These results show that taking insurance will increase push % but decrease loss %. This is reasonable. Wonderful!  However, I’d like you to improve these two things:

#1. As Don mentioned, you need to add a surrender option, because this is huge for counters. Additionally, you need to add the % for surrender, because this is tricky but important for Blackjack.

#2. Specify this as the American rules with a dealer hole card.  

Thank you for trying it out! To do an exact comparison you would have to use the same shuffle seed. The shuffle seed determines the shuffle and composition of the shoes. There are shuffle settings below the first long results table. You can set the shuffle seed to any number but it would have to be the same for exact comparison (if the number 0 is selected or the field is blank the seed is random). The seed used for every run is displayed at the very end of all the results and tables. 

I don't think that taking insurance should change the push %. The hands and insurance bets are evaluated separately.

Phil

We have discussed the random seed issue but I still haven't got the details, but as the rounds of simulation approach many millions, they should not matter very much.  I think. 

I would think so as well.

Can think of the shuffle seed as just one of the many inputs. One that needs to be the same for the exact same shoes. For different shoes (the normal case), change it yourself or just let the program randomly pick it. 

It's just there to hopefully make the shuffle random or make it reproducible if testing something very specific or diagnosing a problem.

Are you saying that your win/loss/push % counts did not include the insurance contribution at all? This part needs to be clarified before we make this comparison. 

Insurance is included in the overall $ amounts and % amounts.

And there are % and $ for the insurance outcomes.

Insurance has nothing to do with whether you win, loss, or push the hand.

Phil

1. Yes, work on surrender. It's too important not to have.

2. OK. Got it for the $25. Makes more sense than my setting my first bet in the ramp to >-99 for play-all. Your way is fine.

3. Yes, think about normalizing to 100 hands per hour, then defining what your own personal visit length is.

4. 8,8 vs. A is NOT a reverse index. What you wrote is OK for 8,8 vs. T but not vs. Ace.

Don

I’ve tested another deviation index, 16v10 hit/stand at HiLo TC=+0.  Here are the numbers for (% hand won, % hand lost, $ hand pushed):

(43.476%, 48.014%, 8.509%) without any index,

(43.520%, 48.067%, 8.411%) with the index of hit/stand at TC=+0.

From these numbers, we find that the % of hand pushed decreases when we use the 16v10 hit/stand index.  This is reasonable, because the player will have more hand values of 16 or less, while the dealer's hand values are mostly greater than 17, which make it harder to match to the player's hand value.  From this point of view, this simulator is reliable too. 

I have a question for you. How do you count % hand won? For example, a player holds an A-A pair hand, splits them apart, and wins both parts.  Do you count this A-A pair as one hand or two?

Two hands.

This is what I thought because your win rate numbers are slightly greater than other people's numbers. How about double-down situations?  For example, a player holds a 5-6 hand and doubles down on it, but loses two units in one hand; do you count this as two hands lost or one?  

One hand.

This is a nice piece of SW - congratulations - and thanks for sharing.

Every time I run the simulation - it ends up being a losing run / losing game [over 1 Million hands].

I even tried to tilt the parameters/strategy options in my favor - and still end up losing in both the short and long run.

Which parameters yield a win over the long run with Advantage Play?

What am I missing or doing wrong here?

You didn't use any bet spread. You flat bet!

Don

Actually I did vary the bet - from $10 to $150 - I didn't include that information in my initial post as I felt it wouldn't fit [too large a file and too large for the screen]... here's a simulation with all the information including varied bets and over 1 Million hands it doesn't get better.

@Don - thanks for your feedback. Agreed varying the bet is clearly Key.

Actually I did vary the bet - from $10 to $150 [unless I'm not understanding the input parameters]

- I didn't include that information in my initial post as I felt it wouldn't fit [too large a file and too large for the screen]... here's a simulation with all the information including varied bets and over 1 Million hands it doesn't get better.

I’ve tested this simulator for two years, so let me try to tell you. If you use HiLo, you need to start to raise your bet at TC=+1, but you set it as TC=+2 and thus missed most of the advantageous betting opportunities.

I still have a question about these win/loss/push percentage numbers. With your simulator, I mostly got 43.5%/48%/8.5%; however, the Wizard’s numbers are often 42.5/49%/8.5%. Is there a discrepancy here?