AutomaticMonkey: CVData 4 now fully functional for SP21

[AutomaticMonkey]

I believe the last bug has been fixed and the QFIT CVData version 4.0106 is now working as well as possible for the S17 variant of Spanish 21.

Here are my results:
1 billion round sim, heads up, 8D, S17, flat bet, Shackleford's CD strategy published on The Wizard of Odds was used:

House edge (IBA%)= -0.418% SD= 1.151 bets/hand.

Compare this to MGP's CA result of -0.373529465859687%

Obviously the CA is a little bit off. (Just kidding MGP!)

The 0.045% difference I suspect may be due to the follow two plays CVData cannot handle- the suit-aware play strategy of hitting a spaded 8,7 vs 4, and the $5000 super bonus and envy bonus. Strategy calls for hitting a suited 77 vs 7, but in the programmed basic strategy this is a split play and there is no way to deviate due to your pair being suited.

Quirks and hacks:

The index generator does not like the special bonuses. To generate indices for whatever count you are using, create a separate sim with no bonuses just for generating indices, then transfer them over to the real sim. Generate a full report with your indices, because they will have to be adjusted a bit for the bonuses. The added benefit of hitting can be 1% to 10%, whether the hit takes advantage of just the push-on-21 rule (multiply the probability of you drawing a 21 by that of the dealer drawing a 21 to get the added advantage) or a multi-card 21 or 6-7-8 bonus.

The index generator does not support multi-card double or the DDR rule. The multi-card double rule makes you less likely to double a play at a given count (because the card you get might give you an even better double hand) and the DDR makes you more likely to double against 8-A, being you can pull half your bet back if you don't like what you get. You're on your own with generating these indices, maybe generating them straight then tweaking them by trial and error will do it. You can judge the value of an index by running two players, one using an index and the other not using it (or using a different one) and compare their win rates for that hand on the Win Rate table.

The automatic win on blackjack rule may or may not be working, kind of confusing, but you can check to see if it working by looking on the data table Advantages and making sure your advantage is always 150% when you are dealt AT, even when the dealer has an ace or 10. I fixed mine by adding an extra 50% bonus whenever the player and the dealer have a 2-card 21 on an unsplit hand.

The included Shackleford strategy for Spanish 21 might be a little bit off. Go ahead and create your own, it's part of the learning experience. You'll need 5 separate strategy tables and the same one is used for 6 card and >6 card hands.

[Norm Wattenberger]

A few notes:

Remember that a CA doesn't take into account cut-card effect. So, -0.373529465859687% may be exactly accurate for fixed rounds -- but 8 decks isn't dealt that way.

The automatic win on blackjack rule looks fine to me. I forgot to turn it on in the SP21 template.

The index generator handles bonuses for CD indexes only. I was going to add support for TD indexes in the last release -- but there has never been a call for this before. I'll put it back on the list.

Thanks for all the work.

> I believe the last bug has been fixed and the QFIT
> CVData version 4.0106 is now working as well as
> possible for the S17 variant of Spanish 21.

> Here are my results:
> 1 billion round sim, heads up, 8D, S17, flat bet,
> Shackleford's CD strategy published on The Wizard of
> Odds was used:

> House edge (IBA%)= -0.418% SD= 1.151 bets/hand.

> Compare this to MGP's CA result of -0.373529465859687%

> Obviously the CA is a little bit off. (Just kidding
> MGP!)

> The 0.045% difference I suspect may be due to the
> follow two plays CVData cannot handle- the suit-aware
> play strategy of hitting a spaded 8,7 vs 4, and the
> $5000 super bonus and envy bonus. Strategy calls for
> hitting a suited 77 vs 7, but in the programmed basic
> strategy this is a split play and there is no way to
> deviate due to your pair being suited.

> Quirks and hacks:

> The index generator does not like the special bonuses.
> To generate indices for whatever count you are using,
> create a separate sim with no bonuses just for
> generating indices, then transfer them over to the
> real sim. Generate a full report with your indices,
> because they will have to be adjusted a bit for the
> bonuses. The added benefit of hitting can be 1% to
> 10%, whether the hit takes advantage of just the
> push-on-21 rule (multiply the probability of you
> drawing a 21 by that of the dealer drawing a 21 to get
> the added advantage) or a multi-card 21 or 6-7-8
> bonus.

> The index generator does not support multi-card double
> or the DDR rule. The multi-card double rule makes you
> less likely to double a play at a given count (because
> the card you get might give you an even better double
> hand) and the DDR makes you more likely to double
> against 8-A, being you can pull half your bet back if
> you don't like what you get. You're on your own with
> generating these indices, maybe generating them
> straight then tweaking them by trial and error will do
> it. You can judge the value of an index by running two
> players, one using an index and the other not using it
> (or using a different one) and compare their win rates
> for that hand on the Win Rate table.

> The automatic win on blackjack rule may or may not be
> working, kind of confusing, but you can check to see
> if it working by looking on the data table Advantages
> and making sure your advantage is always 150% when you
> are dealt AT, even when the dealer has an ace or 10. I
> fixed mine by adding an extra 50% bonus whenever the
> player and the dealer have a 2-card 21 on an unsplit
> hand.

> The included Shackleford strategy for Spanish 21 might
> be a little bit off. Go ahead and create your own,
> it's part of the learning experience. You'll need 5
> separate strategy tables and the same one is used for
> 6 card and >6 card hands.

[jblaze]

Can I just click load -> spanish21.bon? This file seems to add all the bonuses. Also, do you then have to use a modified shoe, or is that already set?

> I believe the last bug has been fixed and the QFIT
> CVData version 4.0106 is now working as well as
> possible for the S17 variant of Spanish 21.

> Here are my results:
> 1 billion round sim, heads up, 8D, S17, flat bet,
> Shackleford's CD strategy published on The Wizard of
> Odds was used:

> House edge (IBA%)= -0.418% SD= 1.151 bets/hand.

> Compare this to MGP's CA result of -0.373529465859687%

> Obviously the CA is a little bit off. (Just kidding
> MGP!)

> The 0.045% difference I suspect may be due to the
> follow two plays CVData cannot handle- the suit-aware
> play strategy of hitting a spaded 8,7 vs 4, and the
> $5000 super bonus and envy bonus. Strategy calls for
> hitting a suited 77 vs 7, but in the programmed basic
> strategy this is a split play and there is no way to
> deviate due to your pair being suited.

> Quirks and hacks:

> The index generator does not like the special bonuses.
> To generate indices for whatever count you are using,
> create a separate sim with no bonuses just for
> generating indices, then transfer them over to the
> real sim. Generate a full report with your indices,
> because they will have to be adjusted a bit for the
> bonuses. The added benefit of hitting can be 1% to
> 10%, whether the hit takes advantage of just the
> push-on-21 rule (multiply the probability of you
> drawing a 21 by that of the dealer drawing a 21 to get
> the added advantage) or a multi-card 21 or 6-7-8
> bonus.

> The index generator does not support multi-card double
> or the DDR rule. The multi-card double rule makes you
> less likely to double a play at a given count (because
> the card you get might give you an even better double
> hand) and the DDR makes you more likely to double
> against 8-A, being you can pull half your bet back if
> you don't like what you get. You're on your own with
> generating these indices, maybe generating them
> straight then tweaking them by trial and error will do
> it. You can judge the value of an index by running two
> players, one using an index and the other not using it
> (or using a different one) and compare their win rates
> for that hand on the Win Rate table.

> The automatic win on blackjack rule may or may not be
> working, kind of confusing, but you can check to see
> if it working by looking on the data table Advantages
> and making sure your advantage is always 150% when you
> are dealt AT, even when the dealer has an ace or 10. I
> fixed mine by adding an extra 50% bonus whenever the
> player and the dealer have a 2-card 21 on an unsplit
> hand.

> The included Shackleford strategy for Spanish 21 might
> be a little bit off. Go ahead and create your own,
> it's part of the learning experience. You'll need 5
> separate strategy tables and the same one is used for
> 6 card and >6 card hands.

[Norm Wattenberger]

That will load bonuses; but not modify the deck or set other crazy rules. Easier to create a new configuration based on the Spanish 21 Template.

> Can I just click load -> spanish21.bon? This file
> seems to add all the bonuses. Also, do you then have
> to use a modified shoe, or is that already set?


Serious Blackjack Software

[PCF]

[Norm Wattenberger]

[PCF]

[Daily Dubble]

> Quirks and hacks:

> The index generator does not like the special bonuses.
> To generate indices for whatever count you are using,
> create a separate sim with no bonuses just for
> generating indices, then transfer them over to the
> real sim. Generate a full report with your indices,
> because they will have to be adjusted a bit for the
> bonuses. The added benefit of hitting can be 1% to
> 10%, whether the hit takes advantage of just the
> push-on-21 rule (multiply the probability of you
> drawing a 21 by that of the dealer drawing a 21 to get
> the added advantage) or a multi-card 21 or 6-7-8
> bonus.

AM:

Can you explain more about what you did to generate indices? I am playing around with a SP21 version of KO and when I eliminate the bonuses, the indices are way too low. How did you "adjust" them for the bonuses? Do you mean just fussing around in a sim using trial and error or am I missing something?

[21forme]

> AM:

> Can you explain more about what you did to generate
> indices? I am playing around with a SP21 version of KO


I use KO for BJ and would have liked to use it for Sp21. However, the EOR of the 7 is 0.01%, per MGP, so counting it as a +1 is not a good idea.

[Daily Dubble]

> I use KO for BJ and would have liked to use it for
> Sp21. However, the EOR of the 7 is 0.01%, per MGP, so
> counting it as a +1 is not a good idea.

Thanks, and you are correct...I have been playing around with trying to develop a modified KO system to work with SP21. I am using the KO system that we already know with only one modification - don't count 7's. Since there are already no 10's, and 7's are not that important in SP21, it seemed like a natural thing to try. The result is an unbalanced count with the same expected RC per deck as for KO for BJ.

After doing some rough conversions of Kat count TCs to my SP21 KO, I simmed with CVDATA and produced similar SCOREs for play all H17 6D (vs. the Kat count).

Using an IRC of zero, the player gets an advantage at a SP21 KO RC=16 and the advantage reaches 1.25% at the pivot point of SP21 KO RC=24 (which corresponds exactly to a Kat count of 0) and 2.75% at an RC of 36.

I added Kat's "illustrious 18" as she describes them on page 91 (Many of these are different than the BJ I18) and converted the Kat count indices to the SP21 KO and refined using CVDATA.

One thing about using an SP21 modified KO count, is that the count does not go negative very often or for very long. You will be counting mostly positive numbers.

I am now working on indices...