Sim Question

[Cohiba]

Norm:

Recently ran a sim with a bet ramp of 1 hand times x the bet for nuetral and negative counts and went to 2 hands times a higher bet at higher counts. After running the sim compared the results to the BJA Pro Betting Software using the same bet ramp and variation between one and two hands and the results were much different. The CVData simulation gave much lower results.

Then I adjusted the bet ramp in CVData to always play one hand and increased the bet size to match the total bet sizes (spread over the two hands) in the BJA Pro Betting software simulation and the results of the sim (ev anyhow variance was higher as expected) matched the output of the BJA Pro Betting software.

My question is whether you understand what is going on? Perhaps its some type of impact associated with spreading to two hands in the CVData as its an actual sim and there are impacts of adding a second hand due to the card eating effect.

Any ideas - thanks.

Cohiba

[Norm]

I asked Colin where the data from BJA Pro cam e from and he said he didn't know. Card eating is a part of the difference in one to two hands switch. The major part is effect on variance. From the CVData results, you can click on Call CVCX on the summary page to see the correct optimal betting results.

[Cohiba]

Thanks Norm - i am noticing different results when i run a CVData sim with two cases and the first case being 1) two hands and a bet spread vs the second case 2) one hand with the same ramp but double the bet size that was used for the two bet sims. This results in the same amount of money bet at each count, but the results are different when reported on an hourly basis.

My assumption is that the rounds or hands played are being treated differently in the two cases. In other words, if I play one hand per round then after 100 rounds I have played 100 hands. But if i play two hands per round then i reach 100 hands played much sooner than in the one hand case. I assume that the reporting is for hands played and therefore the number of rounds played in an hour is different for cases 1 and 2?

Any insights? = thanks.

Cohiba

[Norm]

With one hand you have the normal variance. With two, you have co-variance between two hands against the same dealer hand. You also have some difference in rounds/shoe and you have a change in pushes per round and a small real penetration difference and additional info part way through the round. Everything in blackjack has an effect.

[Cohiba]

Thanks norm. All makes sense. You mentioned a difference in rounds/shoe. When you determine hourly win rates etc per number of assumed rounds played per hour are you actually counting the number or hands played or the number of rounds? For example if I play one round I play one hand per round but if I play two hands I am playing two hands per round. Which are you using - I.e hands played or number of rounds played independent of the number of hands to arrive at hourly totals?

Cohiba

[Norm]

It's always rounds. Rounds are what counts for purposes of risk calculations.

[Cohiba]

Ok so if I assume 100 rounds played per hour and 1 hand played per round then I play 100 hands per hour. On the other hand if I play 2 hands per round and have the same 100 rounds per hour assumption then I am actually playing 200 hands per hour?

Thanks

Cohiba

[Norm]

Yes. Which means you should actually reduce the rounds/hour setting. Actually, a good player can play two hands nearly as quickly as one. But, there will be more shuffles and the dealer may or may not be speedy at dealing and payoffs. I have always maintained that the user must set this number as there is no way a simulator knows the speed of the dealer or other players.

[Cohiba]

Got it. Thanks for the response and for adding clarity.

[Softhony]

With one hand you have the normal variance. With two, you have co-variance between two hands against the same dealer hand. You also have some difference in rounds/shoe and you have a change in pushes per round and a small real penetration difference and additional info part way through the round. Everything in blackjack has an effect.

Dear Norman, greetings, I want to ask you a question, I ran a simulation in CVCX where from the real count 3 I play two hands. In most of the standard deviations on average it is 1.154 whose variance is 1.33, but from the real account 3 the standard deviation increases to 1.903. Is it possible to deduce the variance of the real account 3 from a single hand taking into account a covariance of 0.48 from 1.903? I have been thinking that it could be 1.903^2/2-0.48 = 1.33 Is this operation correct or am I making a mistake and not considering some additional information as you mention?


Sincerely


Anthony Davalos

[Norm]

Heading off to dinner. I'd say you could make an approximation. It wouldn't be totally accurate as changing number of hands in a round changes all count frequencies. I've played with this and it's not easy to get an accurate value algorithmically. A sim is required.

[Softhony]

Thanks Norm for responding, anyway I'm glad to know that the mathematical relationship is not so far-fetched even though it is only an approximation. At least it helps to interpret that 1.903 is the standard deviation of the round if I'm not mistaken.

Sincerely,

Anthony Davalos

[DSchles]

See the formula at the bottom of page 20 of BJA3.

Don