Thread: zoomie: Alternative count for SP21

1. zoomie: Alternative count for SP21

The table below illustrates some thinking on tailoring a multi-level count to SP21. All I did was divide all EORs by the value for a deuce (.290), then round the results to the nearest whole number. The calculations are shown below. I calculate the correlation coefficient as .997, higher than HiLo and RPC (.956 for each) or the Kat Count (.968). (BTW, keeping the Ace at -2 to stay at Level 2 lowers the correlation to .982, still higher than the others.)

I have two questions: First, shouldn't this count be more accurate than the others, leading to a higher SCORE cet par? Second, can I use the CV products somehow to generate indices for this count? Thanks in advance for your help. Sorry, the post loses some formatting of a table . . .

SP21 EOR - Normalized to .290 - Rounded
0.29 1.000 1
0.40 1.379 1
0.56 1.938 2
0.65 2.245 2
0.40 1.379 1
0.04 0.138 0
-0.21 -0.724 -1
-0.17 -0.586 -1
-0.44 -1.510 -2
-0.44 -1.510 -2
-0.44 -1.510 -2
-0.73 -2.528 -3
Correlation Coefficient 0.997

2. zoomie: Follow-up calculations for BJ

FWIW, I created the same rounded tags for regular BJ, and those have a correlation of .977. RPC is almost identical, .975, while HiLo is lower by a bit, .966.

3. Aruuba: Re: Alternative count for SP21

> The table below illustrates some thinking on tailoring
> a multi-level count to SP21. All I did was divide all
> EORs by the value for a deuce (.290), then round the
> results to the nearest whole number. The calculations
> are shown below. I calculate the correlation
> coefficient as .997, higher than HiLo and RPC (.956
> for each) or the Kat Count (.968). (BTW, keeping the
> Ace at -2 to stay at Level 2 lowers the correlation to
> .982, still higher than the others.)
> Correlation Coefficient 0.997

Wish I could help you with your basic question of generating indexes. Maybe you could help me with one?

For some reason I get a correlation coefficient of 0.98 using your level 3 count? And 0.9502 using the hi-lo against her EOR's in case it helps.

Just curious what I'm likely doing wrong. Do I have to adjust something for number of decks?

Does it matter her EOR's don't sum to 0? Are they supposed to?

Any help appreciated.

4. zoomie: Re: Alternative count for SP21

> For some reason I get a correlation coefficient of
> 0.98 using your level 3 count? And 0.9502 using the
> hi-lo against her EOR's in case it helps.

> Just curious what I'm likely doing wrong. Do I have to
> adjust something for number of decks?

I have double-checked my calculations. I am using the CORREL function in Excel and the SP21 EORs from p. 52, Table 5.1 of Ms. Walker's book. But, maybe I am doing something wrong . . .

5. Aruuba: Re: Alternative count for SP21

> I have double-checked my calculations. I am using the
> CORREL function in Excel and the SP21 EORs from p. 52,
> Table 5.1 of Ms. Walker's book. But, maybe I am doing
> something wrong . . .

Thanks for your reply. I use Lotus and my version doesn't even have a correl function.

So I used something I came across

correlation = A / B, where

A = sum { Ci * Pi } and
B = sqrt ( sum of Ci2 * sum of Pi2 )

I got it from the link I put in the box if I'm doing it right.

No idea why the 2 formulae might come up with different values (I mean I can see they're not the same formula) but it would seem one of them might be the wrong one to use since each seem to say they do the same thing? Anyone have any thoughts on the discrepancy or what each is actually measuring or the methodolgy, assumptions used in measuring the correlation? I'd have to think Excel knows what it's doing!

But using the CORREL function in Excel, and thanks for pointing that out - never knew it had that - I do get your values.

6. Aruuba: Re: Alternative count for SP21

> I have two questions: First, shouldn't this count be
> more accurate than the others, leading to a higher
> SCORE cet par? Second, can I use the CV products
> somehow to generate indices for this count?

Thinking more about it, I think you can create index tables for SP21 in CVCX. But not sure if, thru trial and error, you could find which index might be better. Or if CVData could handle it or not.

Heck - just e-mail the man himself and ask him!

I thought maybe Automatic Monkey had a way of doing this with QFIT products and some of the different counts he was proposing.

7. Katarina Walker: Those EORs are for H17 - is that the game you play?

The EORs in my book are just for the H17 Spanish 21 game.
Is that the game you play - the one with the really high house edge: 0.78%?
The EORs for S17 SP21 are significantly different to the H17 EORs, just like in Blackjack (see page 522 in Blackjack Attack by Don Schlesinger, to see how different the EORs are in Blackjack, just from changing H17 to S17). Same with the redouboing game: the EORs are way different, once again (Aces are more valuable, for one thing).
The "Kat Count" was designed using MGP's S17 EORs, for the S17 game. It is the most optimal 2-level count for the S17 game (it is also unbalanced). Obviously, a 3-level count would be better. I admire your ability to be able to use a 3-level count!

I also use the CORREL function in Excel. CORREL equals "betting efficiency", but not the playing efficiency. You want both to be high. Obviously, to get the highest possible betting efficiency, you keep tweaking the tags in Excel until you get the highest CORREL value.

The H17 EORs in my book are actually Wild Bill's. I borrowed them from his BJ21.com post from years ago, in the archives. They should add up to zero, but are slightly off. I think he did them using Monte Carlo simulation, rather than combinatorial analysis (the latter is the more efficient and accurate way to do it. That said, I also use Monte Carlo).
So I believe that the reason why they are slightly off is because not enough sims were performed to get that kind of accuracy. Not to detract from WIld Bill's work. It was revolutionary for its time. Unfortunately, he did it before the S17 Sp21 and redoubling Sp21 games came along, so it wasn't competitive with the good Blackjack games.

With regards to S17 and H17+redoubling SP21, MGP has done the EORs. Ask him to repost them. At the moment, there is still not 100% agreement on the SP21 EORs - it's still in the "R & D" phase, but it's getting there. (Wild Bill's H17 EORs are slightly different to MGP's EORs. But MGP is refining his CA, so I'd put my money on his. He is very good at what he does.) When I have time, I will do them myself and see how they compare with MGP and WIlld Bill.

8. AutomaticMonkey: Actual EOR's might not add up to zero

You mention your EOR's (or preferably, E'sOR!) don't add up to zero and I don't think they're supposed to. The low cards always have a slightly higher sum EOR and I believe that's what leads to the floating advantage in BJ, and SP21 probably has a floating advantage too.

9. zoomie: I wasn't tempted to use new tags

> The EORs in my book are just for the H17 Spanish 21
> game.
> Is that the game you play - the one with the really
> high house edge: 0.78%?

No, and I was not tempted to try using the new tags until I thought I knew what I was doing, which will not be soon . . .

> use a 3-level count!

I doubt that I want to try that. I will do the smart thing - take your advice. Play HiLo until I have indices for the Kat Count! Thanks again.

Posting Permissions

• You may not post new threads
• You may not post replies
• You may not post attachments
• You may not edit your posts
•