Norm Wattenberger: A pile of Insurance Cover sims

[Norm Wattenberger]

The Question

Insurance can be a give-away that you are a counter. What are the costs of various Insurance Cover strategies?

The Sims

There are six types of sims varying the play of Insurance as follows::

[*]Insure all Blackjacks (the ploppy method) [*]No Insurance[*]Do not Insure stiffs, always Insure BJ, correct index otherwise (heavy cover)[*]Increase the index for stiffs to +8, always Insure BJ (light cover)[*]Correct Index (normal counter)[*]Insurance Side Count (max)[/list]

I also ran two sets of these sims as follows:


[*]Single deck, S17 using Zen Ill18, 3:1 spread[*]Six deck, S17 using Zen Ill18, 10:1 spread[/list]

For each set, the following penetrations were simmed:


[*] Single deck: all penetrations from 26 to 39 cards cut off[*]Six deck: all penetrations from 26 to 104 cards cut off[/list]



The Results

Note: The normal Insurance indexes for Zen are +4 for single deck and +5 for six decks. Increasing the indexes to +8 for stiffs roughly halves these Insurance bets. The links below display two charts for single and six decks. These are area charts. The SCOREs are plotted for each penetration. Insuring a BJ only (the ploppy strategy) is used as a base. Successively better strategies are added to the charts with different colors indicating the improvement gained from each strategy. The narrow pink area is the difference between using the correct index and the simmed light cover. These charts indicate that some Insurance cover is acceptable. (Apologies for the colors - some resolution is lost in posting.)


[*] Single deck chart[*]Six deck chart[/list]



Sim details

All sims were 1,000 million rounds, one player, quarter deck resolution, original Zen Illustrious 18, no errors. Optimal betting ramps were calculated for every penetration.

[Don Schlesinger]

Beautiful piece of work, Norm. High-quality stuff. I was actually particularly impressed that you would write "1,000 million" at the end, rather than "one billion," as if to indicate that you realized that some Europeans would be reading this, for whom "a billion" is our trillion! Was that actually your intention?

One nit-pick. In your legends, for stiffs, did you mean to write 12-16, instead of 13-16?

Don

P.S. From where did you get the +4 SD insurance index for Zen? JA gives +3 in his Systems 101, in BJRM. I'm not a Zen player, so I'm not sure which is more accurate.

[Norm Wattenberger]

I was actually particularly impressed that you would write "1,000 million" at the end, rather than "one billion," as if to indicate that you realized that some Europeans would be reading this, for whom "a billion" is our trillion! Was that actually your intention?

The blame is on the French. They changed the meaning of billion, trillion, etc. and we took their definitions instead of the original English meanings. Yes that was in my mind as well the fact that people don't see the "b" in billion.

One nit-pick. In your legends, for stiffs, did you mean to write 12-16, instead of 13-16?

Actually, my error was in using the word stiff. I simmed 13-16. CVData handles 21 through 13 separately but 12 and lower together for purposes of Insurance cover. I think I'll change that.

P.S. From where did you get the +4 SD insurance index for Zen? JA gives +3 in his Systems 101, in BJRM. I'm not a Zen player, so I'm not sure which is more accurate.

+4 is in the original Blackbelt in BJ. As you know, TC frequencies tend to jump around a lot in single deck. There aren't a large number of hands at +3 making the difference insignificant.

[Don Schlesinger]

> The blame is on the French.
> They changed the meaning of billion,
> trillion, etc. and we took their definitions
> instead of the original English meanings.
> Yes that was in my mind as well the fact
> that people don't see the "b" in
> billion.

Well, the French have a separate word for our "billion," which is "milliard," whereas they also have a word, "billion," which is now our trillion.

But, 10^12, meaning billion, is given in my dictionary as "mainly British," so I'm not sure exactly who is to blame here.

> I simmed 13-16. CVData handles
> 21 through 13 separately but 12 and lower
> together for purposes of Insurance cover. I
> think I'll change that.

Yes, I think you should.

Don

[Contrail]

> P.S. From where did you get the +4 SD
> insurance index for Zen? JA gives +3 in his
> Systems 101, in BJRM. I'm not a Zen player,
> so I'm not sure which is more accurate.

for SD (65.4 % pen.) with the parameter for specifying the confidence in the standard error set to 4.

[Don Schlesinger]

> for SD (65.4 % pen.) with the parameter for
> specifying the confidence in the standard
> error set to 4.

Kinda weird, since JA used SBA to derive the indices he catalogued in Systems 101.

John, are you there?

Don

[John Auston]

> Kinda weird, since JA used SBA to derive the
> indices he catalogued in Systems 101.

> John, are you there?

I probably truncated, when I should have rounded.
Might also not have run enough trials.

I bet the diff is small, though.

I'll change it, in future releases.

John

[Don Schlesinger]

> I'll change it, in future releases.

I wouldn't change one index, unilaterally, if you have used a particular methodology to generate all of your indices for Systems 101.

Better to explain, upfront, what that methodology is, rather than tweak an occasional number here and there, unless, of course, this wasn't your methodolgy!

Don

[Richard Reid]

Norm:

Very nice job!

I want to make sure I understand the 6 deck chart correctly. Can we extrapolate the chart and say that with about 1.5 decks cut off in a 6 deck game, if we only take insurance on Blackjacks, there will be very little difference in our SCORE compared with properly insuring all hands when appropriate?

Sincerely,
Richard Reid

[Cacarulo]

> I want to make sure I understand the 6 deck
> chart correctly. Can we extrapolate the
> chart and say that with about 1.5 decks cut
> off in a 6 deck game, if we only take
> insurance on Blackjacks, there will be very
> little difference in our SCORE compared with
> properly insuring all hands when
> appropriate?

Norm:

Is there any way to see what those differences in SCORE are? Perhaps a different chart.

Sincerely,
Cacarulo

[Don Schlesinger]

> Is there any way to see what those
> differences in SCORE are? Perhaps a
> different chart.

A tabular array, at, say, just the half-deck increments, for 6-deck, might be nice.

Don

[Norm Wattenberger]

Not as easy to see as a chart:

Raw data

[Cacarulo]

Thanks Norm, very interesting!

Sincerely,
Cacarulo