jblaze: Multi parameter indices

[jblaze]

Just wondering how you come up with indices for a sidecount of 7s, say. Is there a CV product that does that? CVData?

[Myooligan]

> Just wondering how you come up with indices for a
> sidecount of 7s, say. Is there a CV product that does
> that? CVData?

I don't know if CV will do it for you automatically, but here's how to do it manually (It's actually going to be easier than it might sound, though it will take a bit of monotonous data entry on your part):

First off, get a set of playing eors, such as the ones in BJA3 or Theory of Blackjack. Identify play decisions where 7s have large eors. Next, construct an "alternative count" for the given decision. For instance, for HiLo, 2-6 and 10-A will remain at +1 and -1, but 7s might be +3. The value of the seven will obviously vary by play; this value will be the "multiplier:" you will multiply that number times the "running side count" and add it to your RC.

How do you determine the proper number to use as a multiplier? You need to determine the value that maximizes the correlation coefficient for the given play decision. It's fairly easy; I'll post the formula here if you want.

After determining the proper multiplier value, you can use CVData or SBA to generate a new index number for that particular decision, using your "alternate count" values. Then you repeat the process with each index where the seven is important.

You may have noticed that these directions leave you with a mixture of balanced and unbalanced alternate counts. There are two ways to deal with this, and I'll be happy to say more about them. . . if you're still with me at this point. Just let me know.

BTW, I developed a spreadsheet that does the first half of the work for you (spits out the multipliers for each relevant decision). It's currently getting a tune up at Schlesinger's BJ Analysis Repair Shop, and some version of it may show up here on ap.com one of these days. . .

[jblaze]

I guess before I go through all this, is it worth it? How much will it improve playing efficiency for UAPC? In only positive counts? In shoe games? In pitch games? Essentially, is it worth it?

> I don't know if CV will do it for you automatically,
> but here's how to do it manually (It's actually going
> to be easier than it might sound, though it will take
> a bit of monotonous data entry on your part):

> First off, get a set of playing eors, such as the ones
> in BJA3 or Theory of Blackjack. Identify play
> decisions where 7s have large eors. Next, construct an
> "alternative count" for the given decision.
> For instance, for HiLo, 2-6 and 10-A will remain at +1
> and -1, but 7s might be +3. The value of the seven
> will obviously vary by play; this value will be the
> "multiplier:" you will multiply that number
> times the "running side count" and add it to
> your RC.

> How do you determine the proper number to use as a
> multiplier? You need to determine the value that
> maximizes the correlation coefficient for the given
> play decision. It's fairly easy; I'll post the formula
> here if you want.

> After determining the proper multiplier value, you can
> use CVData or SBA to generate a new index number for
> that particular decision, using your "alternate
> count" values. Then you repeat the process with
> each index where the seven is important.

> You may have noticed that these directions leave you
> with a mixture of balanced and unbalanced alternate
> counts. There are two ways to deal with this, and I'll
> be happy to say more about them. . . if you're still
> with me at this point. Just let me know.

> BTW, I developed a spreadsheet that does the first
> half of the work for you (spits out the multipliers
> for each relevant decision). It's currently getting a
> tune up at Schlesinger's BJ Analysis Repair Shop, and
> some version of it may show up here on ap.com one of
> these days. . .

[AutomaticMonkey]

> I guess before I go through all this, is it worth it?
> How much will it improve playing efficiency for UAPC?
> In only positive counts? In shoe games? In pitch
> games? Essentially, is it worth it?

In a word, no. Especially in a shoe game, where playing efficiency doesn't buy you all that much anyway. If I'm going to sidecount, it's going to be for some lucrative sidebet rather than trying to squeeze a few pennies more out of the BJ part.

[jblaze]

Thanks, that's what I kind of figured. (I posted earlier about super 7s too but no one responded.)

> In a word, no. Especially in a shoe game, where
> playing efficiency doesn't buy you all that much
> anyway. If I'm going to sidecount, it's going to be
> for some lucrative sidebet rather than trying to
> squeeze a few pennies more out of the BJ part.

[Norm Wattenberger]

CVData and CVBJ support multi-parameter tables. But after simming MP; the results were so disappointing I put MP index generation way down the list. I'll probably add it one day for completeness. But as Myooligan says, you can trick CVData into creating them with a feew calculations.

[ToAnyOne]

... super 7 side bet, you could get the majority of the gain potential from the multi-parameter tables by using this information to decide when to hit or stand on 14v10; every 7 is worth several hi-lo points. If there is 1 1/2 missing 7's(approx.) per deck you stand, you can pretty much ignore the primary count for this decision.

Side note: if you counted 5's as well, you would have situations where you would hit a 16v10 but stand on your 14 against the same 10.