Ed Tice: CVData and Insurance SC

[Ed Tice]

I think I am beating the insurance side count to death, but I have a lot of free time on my hands and not a lot of casinos around, so I spend about 95% of my time practicing and 5% playing and I'm always trying to learn new things.

I have gotten fairly proficient (translate: not good enough to play in the casino) with keeping the side count. But I am still not convinced exactly how much it is worth to me. According to PBJ it is worth an extra $2 per hour when spreading 1-10 on a double deck game (not much). But I'd like to know exactly what it is worth to me under different casino conditions. The ability to plug in my own conditions is why I buy simulation software!

When I run my simulation using Hi/Lo and insure at +3 I get one set of data for particular conditions. If I go to the rules screen and uncheck insurance, I get another (lower EV) set of numbers. Now the kicker is that when I reenable insurance AND select "insurance SC" in the playing options, I get the same results as when insurance is not offered. So my conclusion is that I have something configured wrong and that setting the insurance SC is resulting in my *never* taking insurance, even in favorable opportunities.

The other possibility is, of course, that there never *are* favorable opportunities, but if this were true, taking insurance with Hi/Lo TC +3 would not yield any increased EV. (My sims are long enough that I didn't just 'get lucky' with some 'wrong' insurance bets).

Any ideas here? Should I try exporting again?

Also is there a repository of configurations for particular common casino/betting situations (and expected results from running them!) that we can download to confirm that our sims are working correctly?

Ed

[Norm Wattenberger]

Use the Tools - Export function and send me the file

[Dancer]

> I have gotten fairly proficient (translate:
> not good enough to play in the casino) with
> keeping the side count. But I am still not
> convinced exactly how much it is worth to
> me. According to PBJ it is worth an extra $2
> per hour when spreading 1-10 on a double
> deck game (not much).

Ed,

What unit size is PBJ talking about here? I did an insurance study about a year ago and found that perfect insurance gained the Hi-Lo player .093 units/hour in a double deck game assuming the average bet was 3 units (1.5 for insurance) when the bet was made. In 6 deck, it gained the player .083 units/hour assuming the average bet was 5 units (Don S. suggested 5.26 in BJA, I believe) (2.5 for insurance) when the bet was made.

That being said, a black chip Hi-Lo player would gain about $9.30/hour in a double deck game with perfect insurance.

[Ed Tice]

I don't want to republish too much of PBJ here since the copyright has not expired. So if anybody thinks I have crossed the line, just say so and I'll bust my post.

But PBJ defines a reference game of a six deck shoe five dealt out.

S17 DOA, NDAS, Split to 4 hands, No RSA, no surrender, high/low counting with indices from -10 to +10.

Spreading $10 - $100 yields a win rate of $16. Adding perfect insurance makes this $18 per hour. So you win about 10% more money. Standard deviation and other information is also published but you'll have to buy the book for that.

The other statistic that is published is that you will only win $14/hr if you never take insurance.

So the difference between no insurance and perfect insurance is over 25% for a Hi/Lo player. Of course I currently take my insurance at +3 so I'm only gaining the extra $2 if I could do the perfect insurance SC under casino conditions (which I can't do yet.)

Since we are having the discussion of the insurance SC, I'll bring up another topic (although it probably should be moved to another thread and I'll bring it up there if nobody bites).

The count as defined by Wong is IMHO hard to keep. It is easier for me to count two separate side-counts (such as 7s and 9s or Aces and 7s) than it is to keep this count. Therefore, I have thought of keeping a running count of the number of 10s left in the deck/shoe and a number of cards left. It would give me exact card resolution for decisions (negligible) but would mean I never had to look at the discard tray. Also the insurance decision would be trivial... multiply the number of 10s left by 3... if its more than the number of cards left, take insurance... in all but the closes situations this math woudl be trivial. Am I missing something?

Ed

> Ed,

> What unit size is PBJ talking about here? I
> did an insurance study about a year ago and
> found that perfect insurance gained the
> Hi-Lo player .093 units/hour in a double
> deck game assuming the average bet was 3
> units (1.5 for insurance) when the bet was
> made. In 6 deck, it gained the player .083
> units/hour assuming the average bet was 5
> units (Don S. suggested 5.26 in BJA, I
> believe) (2.5 for insurance) when the bet
> was made.

> That being said, a black chip Hi-Lo player
> would gain about $9.30/hour in a double deck
> game with perfect insurance.

[Dancer]

> But PBJ defines a reference game of a six
> deck shoe five dealt out.

> S17 DOA, NDAS, Split to 4 hands, No RSA, no
> surrender, high/low counting with indices
> from -10 to +10.

> Spreading $10 - $100 yields a win rate of
> $16. Adding perfect insurance makes this $18
> per hour. So you win about 10% more money.
> Standard deviation and other information is
> also published but you'll have to buy the
> book for that.

That's a bit higher than the 8.3% I found in my study but certainly in the ballpark.

> The count as defined by Wong is IMHO hard to
> keep. It is easier for me to count two
> separate side-counts (such as 7s and 9s or
> Aces and 7s) than it is to keep this count.
> Therefore, I have thought of keeping a
> running count of the number of 10s left in
> the deck/shoe and a number of cards left. It
> would give me exact card resolution for
> decisions (negligible) but would mean I
> never had to look at the discard tray. Also
> the insurance decision would be trivial...
> multiply the number of 10s left by 3... if
> its more than the number of cards left, take
> insurance... in all but the closes
> situations this math woudl be trivial. Am I
> missing something?

If you can count Hi-Lo while simultaneously keeping 2 independent decrementing counts beginning at 312 and 96, respectively, constantly divide the 312 count by 52 to resolve the TC, flirt with the cocktail waitress, talk to the pit boss about Saturday night's fight and the guy next to you who just bought a new dog -- flawlessly, IMHO, you're way to smart to be wasting your time playing blackjack.

Curing cancer has got to be easier.

Seriously, if you're that good with numbers, why not just learn a more powerful count? According to Don's BJA, you can pickup that same additional 8% - 10% or so just by moving to Hi-Opt II. It only requires 1 side count -- that starts at 0. Piece of cake...

[Magician]

> Spreading $10 - $100 yields a win rate of
> $16. Adding perfect insurance makes this $18
> per hour. So you win about 10% more money.

Note that the betting ramp in PBJ is not optimal. This could explain the difference with Dancer's results.

> The count as defined by Wong is IMHO hard to
> keep.

I don't have the first edition of PBJ so I'm not sure how Wong defined it, but the Insurance count comes in two forms - balanced (tens -9, others +4, bet at TC>4) and unbalanced (IRC -4xdecks, tens -2, others +1, bet at RC>0). The unbalanced form is much easier to use.

> Therefore, I have thought of keeping a
> running count of the number of 10s left in
> the deck/shoe and a number of cards left.

Sounds like the Thorp 10 count. Most would say that there are easier counts that will win more.

[Ed Tice]

> If you can count Hi-Lo while simultaneously
> keeping 2 independent decrementing counts
> beginning at 312 and 96, respectively,
> constantly divide the 312 count by 52 to
> resolve the TC, flirt with the cocktail
> waitress, talk to the pit boss about
> Saturday night's fight and the guy next to
> you who just bought a new dog -- flawlessly
> , IMHO, you're way to smart to be wasting
> your time playing blackjack.

I can keep my Hi/Lo without any trouble at this point, regardless of the distractions. But I don't think I have any aptitude for it. I just practice. A LOT. Like 2-3 hours every night when I don't go to the casino, plus the time I'm in the casino.

I don't want to change counts, because it involves going from something I can do very well to something that I don't do very well. I'd rather learn something additional like shuffle tracking.

I have done a lot of experiments in regards to what side counts I can keep and here are my results:

Counting 3 cards at a time (because this is the easiest way to keep the insurance SC).

Count Nothing (Just Flip the cards): average 12.4 seconds 100% accuracy

Count High/Low: average 12.4 seconds, 100% accuracy

Count High/Low with a side count of aces or sevens: 14.6 seconds, 100% accuracy

Count High/low with two side counts (A,7 or 7,9): 27.8 seconds, 90% accuracy

Count High/Low with the insurance side count: 18.8 seconds, 90% accuracy.

So obviously as hard as the insurance side count it, it is 'easier' (by this metric) than the two side counts. But when I have to count two cards at a time, I can't keep it accurately enough to get a reasonable measure, while I can keep the two side counts at the same speed.

Here is why I think counting down the tens and the cards is easier.

(1) Counting the cards played is fairly easy since it is 2*(No Spots+1) + the hit cards. I can do it very fast at the table. I do this periodically when confirming the shuffle point of a game (to make sure it agrees with my estimates, simulations, etc).

(2) Coutning the 10s down from 96 would be a bit daunting. But counting them up from zero is not so hard. So I can just subtract after each rount... start with 96 tens... nine came out.. so now we have 87....

(3) Division is hard. It is much harder than multiplication. I can do two or three multiplicatoin problems in the time I can do one division problem. The whole point of this way is that ther eis NO DIVISION. You just take the number of unplayed tens and multiply by three... if you get more than the number of cards, insurance is useful.

So 203 cards left, 68 tens left... 68*3 = 180 + 24 = 204 so take insurance (marginal).

The advantage of this technique is that, if you can multiply left-to-right you often dont need to finish the multiplication in order to have the answer. So lets say you have 85 cards left and 31 tens. 30 x 3 = 90 and you odnt need to finish the multiplication.

Keeping the number of cards left is relatively easy in that you can do it before the deal.... if there are six players (including you) you can subtract 14 before the first card is dealt... this is the number of cards that will be left if/when the insurance call is made.... so the only hard part is if the insurance call is made, coutning the number of tens, subtracting and multiplying, but I can do this fast enough.

I have tried keeping this count (although I have not used it for insurance since I didn't have confidence in it).

I normally look away during the deal of the first cards (like at the roulette results as if they were interesting) and then count my Hi/Lo as the second card is flopped.

If the dealer shows an ace, I ignore the Hi/Lo while the second card comes out and just figure out the insurance bet. The biggest downside that I've found is that when the dealer DOES have blackjack I have to get the High/Low count REALLY fast before all the cards get picked up... and I am always worried that it will attract heat (Who but a counter cares about what cards people had when the dealer turns over a board-clearing BJ)

However I've been playing mostly in Europe now with the ENHC rule, which makes the technique much easier. The cards stay on the table much longer, so I can pick up the Hi/Lo when the other players play their hands.

I hope I'm clear here. I'm not trying to disagree with you. I'm challenging your statements only for the purpose of continuing the conversation and stirring up some discussion.

Ed

[Dancer]

> I can keep my Hi/Lo without any trouble at
> this point, regardless of the distractions.
> But I don't think I have any aptitude for
> it. I just practice. A LOT. Like 2-3 hours
> every night when I don't go to the casino,
> plus the time I'm in the casino.

> I don't want to change counts, because it
> involves going from something I can do very
> well to something that I don't do very well.
> I'd rather learn something additional like
> shuffle tracking.

> I have done a lot of experiments in regards
> to what side counts I can keep and here are
> my results:

> Counting 3 cards at a time (because this is
> the easiest way to keep the insurance SC).

> Count Nothing (Just Flip the cards): average
> 12.4 seconds 100% accuracy

> Count High/Low: average 12.4 seconds, 100%
> accuracy

> Count High/Low with a side count of aces or
> sevens: 14.6 seconds, 100% accuracy

> Count High/low with two side counts (A,7 or
> 7,9): 27.8 seconds, 90% accuracy

> Count High/Low with the insurance side
> count: 18.8 seconds, 90% accuracy.

> So obviously as hard as the insurance side
> count it, it is 'easier' (by this metric)
> than the two side counts. But when I have to
> count two cards at a time, I can't keep it
> accurately enough to get a reasonable
> measure, while I can keep the two side
> counts at the same speed.

> Here is why I think counting down the tens
> and the cards is easier.

> (1) Counting the cards played is fairly easy
> since it is 2*(No Spots+1) + the hit cards.
> I can do it very fast at the table. I do
> this periodically when confirming the
> shuffle point of a game (to make sure it
> agrees with my estimates, simulations, etc).

> (2) Coutning the 10s down from 96 would be a
> bit daunting. But counting them up from zero
> is not so hard. So I can just subtract after
> each rount... start with 96 tens... nine
> came out.. so now we have 87....

> (3) Division is hard. It is much harder than
> multiplication. I can do two or three
> multiplicatoin problems in the time I can do
> one division problem. The whole point of
> this way is that ther eis NO DIVISION. You
> just take the number of unplayed tens and
> multiply by three... if you get more than
> the number of cards, insurance is useful.

> So 203 cards left, 68 tens left... 68*3 =
> 180 + 24 = 204 so take insurance (marginal).

> The advantage of this technique is that, if
> you can multiply left-to-right you often
> dont need to finish the multiplication in
> order to have the answer. So lets say you
> have 85 cards left and 31 tens. 30 x 3 = 90
> and you odnt need to finish the
> multiplication.

> Keeping the number of cards left is
> relatively easy in that you can do it before
> the deal.... if there are six players
> (including you) you can subtract 14 before
> the first card is dealt... this is the
> number of cards that will be left if/when
> the insurance call is made.... so the only
> hard part is if the insurance call is made,
> coutning the number of tens, subtracting and
> multiplying, but I can do this fast enough.

> I have tried keeping this count (although I
> have not used it for insurance since I
> didn't have confidence in it).

> I normally look away during the deal of the
> first cards (like at the roulette results as
> if they were interesting) and then count my
> Hi/Lo as the second card is flopped.

> If the dealer shows an ace, I ignore the
> Hi/Lo while the second card comes out and
> just figure out the insurance bet. The
> biggest downside that I've found is that
> when the dealer DOES have blackjack I have
> to get the High/Low count REALLY fast before
> all the cards get picked up... and I am
> always worried that it will attract heat
> (Who but a counter cares about what cards
> people had when the dealer turns over a
> board-clearing BJ)

> However I've been playing mostly in Europe
> now with the ENHC rule, which makes the
> technique much easier. The cards stay on the
> table much longer, so I can pick up the
> Hi/Lo when the other players play their
> hands.

> I hope I'm clear here. I'm not trying to
> disagree with you. I'm challenging your
> statements only for the purpose of
> continuing the conversation and stirring up
> some discussion.

> Ed

You've clearly given this a considerable amount of thought. You're certainly going well beyond the current wisdom of the game that less is more. Only you can answer whether it's all worth it.

Just think, KO is one of the most popular counts because people have too much difficulty just converting to true count -- let alone imagining what you're proposing.

In the end, you're considering extreme measures to pick up a few percentage points of additional profits. IMO, if you want them that badly, those same additional profits -- more in fact (according to a few quick sims with CVData) -- can be garnered for far less effort.

[Ed Tice]

Thanks for the comments. Yes, I realize that I'm talking about a reasonable amount of extra effort here. What I like about it is that I can do all of the calculations while the dealer is paying the last bets/flopping the first cards, while i'm looking away at something else. So I have lots of time.

But what has always appealed to be about the insurance side count is not the few extra $ in EV, but rather that it will tell us when to take insurance even in neutral/negative counts. Some people take insurance when its not called for, as a matter of cover. Some take it always. I'd prefer to take it for 'cover' when its actually a good bet. I realize that the additional EV is not worth the efforts. But taking insurance at negative counts with a minimum bet out as cover while increasing EV is appealing to me.

Thanks for the feedback.

BTW, I am still not doing this in the casino. I practice a LOT of things at home, but when I hit the tables, I guess I am conservative and stick with the trietd and true!

Ed

> You've clearly given this a considerable
> amount of thought. You're certainly going
> well beyond the current wisdom of the game
> that less is more. Only you can answer
> whether it's all worth it.

> Just think, KO is one of the most popular
> counts because people have too much
> difficulty just converting to true count --
> let alone imagining what you're proposing.

> In the end, you're considering extreme
> measures to pick up a few percentage points
> of additional profits. IMO, if you want them
> that badly, those same additional profits --
> more in fact (according to a few quick sims
> with CVData) -- can be garnered for far less
> effort.

[Dancer]

> Thanks for the comments. Yes, I realize that
> I'm talking about a reasonable amount of
> extra effort here. What I like about it is
> that I can do all of the calculations while
> the dealer is paying the last bets/flopping
> the first cards, while i'm looking away at
> something else. So I have lots of time.

> But what has always appealed to be about the
> insurance side count is not the few extra $
> in EV, but rather that it will tell us when
> to take insurance even in neutral/negative
> counts. Some people take insurance when its
> not called for, as a matter of cover. Some
> take it always. I'd prefer to take it for
> 'cover' when its actually a good bet. I
> realize that the additional EV is not worth
> the efforts. But taking insurance at
> negative counts with a minimum bet out as
> cover while increasing EV is appealing to
> me.

> Thanks for the feedback.

> BTW, I am still not doing this in the
> casino. I practice a LOT of things at home,
> but when I hit the tables, I guess I am
> conservative and stick with the trietd and
> true!

> Ed

From a solely cover perspective, I agree. I submitted a note on the Blackjack Main board recently about this exact topic (search for Camo Comments).

The only additional thought I have for your approach is that it only provides cover for your insurance bets. It won't provide any cover for how you play your hands vs. how you bet. Your PE -- even with 2 side counts and the additional math -- is still only 51%. That leaves a fair amount of EV (and a substantial number of cover opportunities) on the table.

[Ed Tice]

Agreed. As I said I don't keep this count. I have experimented with it though, and would like to. I can actually side count 7s and 9s without any trouble and it is MUCH easier than the insurance count (either my way or the traditional way). I use the (7-9)/2 adjustment for betting and playing which gives me BE and PE close to the halves count but still the full IC of Hi/Lo. Plus I adjust my play of 14s vs 10. I play mostly shoes, though, so I dont get too many chances to stand on it. But I love when I do because everybody goes crazy and nobdy thinks its a counter's play.

Speaking of PE, though, do you have any indices for 12vs 10 hitting? One would think that they might be similar to 14v10 that you stand on 12v10 when there is only a certain 9 density. But I am not sure this is true. Even with no 9s, you might hit 12v10 because there is still a chance to draw to something like a 15 and hit again.... of course in high counts you aren't hitting that 15.... so maybe there is such an index... but I can't reason about it logically, I need to come up with some computation and I don't know how to find this out.

Ed

> From a solely cover perspective, I agree. I
> submitted a note on the Blackjack Main board
> recently about this exact topic (search for
> Camo Comments ).

> The only additional thought I have for your
> approach is that it only provides cover for
> your insurance bets. It won't provide any
> cover for how you play your hands vs. how
> you bet. Your PE -- even with 2 side counts
> and the additional math -- is still only
> 51%. That leaves a fair amount of EV (and a
> substantial number of cover opportunities)
> on the table.