insurance @ TC +3? plz. explain

Hi. I've been thinking about this one for quit a while. And i've been trying to find the answer on the message board with the search, but without sucess. I'll be heading out to the local casino in a few hours so if possible, could anyone please respond ASAP, i'd much appreciate it.

Paraphrasing the blackjack school here, it says, "the high/low counting system has an 'insurance efficiency' of 80% which means that 8 out of 10 times you'll be doing the right thing when making an insurance bet based on the true count." (the magic number being 3).

So logically I thought about it. In a deck there are four suits of each card. But when illustrating my thought process, assuming one of each card should be okay (correct?).

For an even count there is:
Ace, king, queen, jack, ten: (5 cards with value of -1)
nine, eight, seven: neutral (3 cards)
six, five, four, three, two: (5 cards with value of +1)

So, to achieve a true count of +3: (lets assume these cards haven't been played, and are still in the shoe)
Ace, king, queen, jack, ten: (5 cards)
nine, eight, seven: neutral (3 cards)
six, five: (2 cards)
**four, three, and two have been already played leading to a TC of +3: (3 cards)**
**Subtracting the four, three, and two which have been played, 10 cards remain.**

so with a dealer Ace showing and a TC of +3 (subtracting for the Ace showing)
possible hole cards:
Ace: (leading to no blackjack) 1 card
King, queen, jack, ten: (leading to a possible blackjack) 4 cards
nine, eight, seven: (3 cards leading to no blackjack)
six, five: (2 cards leading to no blackjack)

i figure the math is like this:
4 out of the remaining 10 cards leads to a blackjack (king, queen, jack, ten)=40%
6 out of the remaining 10 cards leads to a no blackjack (nine, eight, seven, six, five, and Ace)=60%

So i scratch my head and wonder how can there be an "Insurance effieciency of 80%? I'm no mathmatician but i think i'm pretty strong logically. If someone could please explain this to me very simply like I have I would much appreciate it.

Does one have to remove the neutral cards out of the equation altogether? I thought about that, but it didn't make sense to me. I assume they would remain in the deck, and there is no possible way to track them since the high/low system only keeps track of high and low cards, no?

MUCH aprreciated: and respect to all AP out there.

An Insurance Efficiency of 80% does not mean that the count will successfully predict a natural 80% of the time. It means that you will be making an even or favorable bet 80% of the time if you take insurance. In single-deck play, one needs to see a hi-lo TC of only 1.4 or greater to make the insurance bet. To make insurance an even bet or better, the ratio of tens to non-tens only has to be one out of three or less. This is because the insurance bet pays 2:1, which means you only have to win this bet once in three tries to break even. In your scenario where a player has a 40% chance of winning the bet, the EV for the player is +20%. That's pretty hefty.

Thank you for the prompt response bootlegger. I understand what it means now 'insurance efficiency'. I wish the blackjack school explained it in a more simple way, because the way it's worded can be confusing.

So 80% of the time you'll make an even money or favorable bet. Isn't that less than breaking even in the long run then? Shouldn't the efficiency be 100% or am I confused here?

Your explanation about the ratio of tens to non-tens only having to be one out of three or less because of the 2:1 payout was superb, and couldn't be explained better. That I understand very well. But the one thing I still don't understand is, how does the number of TC +3 be the deciding factor (in a six deck game) for taking insurance?

I thought my scenario wasn't special. I just thought I simplified the situation of a TC +3.
much appreciated.

P.S. What does EV stand for?

By the way, I looked it up on Richard Reid's site, and the Insurance Correlation given there for hi-lo is 76%. That's enough to make money on the insurance bet, however. Taking insurance starts at 1.4 for single deck, goes to 2.4 for double deck, and 3 for six deck. Practically speaking, for most TC conversion methods and utilizing truncating, you would do it at +2 for single deck and +3 for multiple deck games. That's where you start. Obviously, there will be many times when you are taking insurance at higher counts, where the advantage is greater. On average, taking insurance at the suggested counts will result in a significant profit for the player.

EV stands for "expected value." It's another way of saying what your expectation is for a given game. It is an essential concept to learn if you're just starting out.

It means that you will be making an even or favorable bet 80% of the time if you take insurance.

I don't think this is correct. 80% IC (or IE which is used interchangably with IC, thought not neccessarily correctly) means the count system gets 80% of the gain from a perfect insurance system. That isn't the same thing as making a positive EV bet 80% of the time you take insurance. By using a higher strike number you can easily make positive EV bets almost 100% of the time when you take insurance, with any responsible commercial system-but you would be missing out on many favourable opportunities also.

With my sim program, I had the high-low and the perfect insurance count running simultaneously.
I found that the insurance decision triggered by high-low (TC+3) was erroneous in about 50% of the cases!

Francis Salmon

"I found that the insurance decision triggered by high-low (TC+3) was erroneous in about 50% of the cases!"

That sounds about right, if you mean the insurance decision at exactly TC +3. An index number is supposed to show the point where a decision changes from unfavorable to favorable. It would be even more support for 3 as the correct Insurance Count if the average of all the 10/others ratios came out to .5

My error.

I did some studies on this a long time ago, and if there is an interest, I can try to dig up the results. I am heavily engaged right now trying to get mid-term grades in, so it may be a while before I can do that. But I will talk some from memory.

Yes Hi-Lo will make a lot of mistakes regarding insurance. However, many of those mistakes will be when your insurance is close play, so the cost of those errors is not great.

When insurance is a big ev play (pack rich in 10s), HiLo will get it right a lot more often. And when insurance is a big sucker play, HiLo will get that right.

Finally, here is an important point that is overlooked in most of the literature. Classical Insurance Efficiency is based upon flat-betting. But most HiLo players are not flat-betting. HiLo is a lot more accurate in the high counts, when the big bets are out.

The IE for a HiLo person spreading their bets with the count is significantly higher than the efficiencies quoted above.

Consider a theoretical situation in which you could grab all the cards remaining in a shoe whenever the TC reaches or exceeds the Hi-Lo insurance index and an insurance decision is required. Do this regardless of the size of the remaining shoe.

Is it correct to say that if you tossed ALL of those cards from ALL of those qualifying shoes into one big pile that at least one-third of the cards in that entire pile would be Tens? I think that's the essence of the index, isn't it? It pinpoints the place on the number-line where this first occurs.

Now, if you had stopped to examine each individual shoe-remainder before tossing it into your big pile of cards, are you saying that ~50% of the time the Tens in an individual shoe constituted less than one-third of the cards in the shoe?

I think both of these are correct, but I'm not positive. It sounds reasonable to me. Wouldn't there be plenty of shoes in which the Tens make up 40%, 50%...even 60% of the remaining cards in a shoe, especially as the count goes higher than the index? For the index to be correct, there would also have to be a significant number of off-setting shoes in the entire galaxy of qualifying shoes which have less than one-third Tens. Otherwise, the index wouldn't be valid and would have to be "moved".

I just wanted to show that the high-low count is not very efficient as far as insurance is concerned and don't pin me down on the 50%-figure. I just cited this from memory but i'm sure it was an astonishingly high percentage.
It includes many cases where the insurance wasn't taken even if it was called for.

Francis Salmon

I think the answer to your question is YES. For any count (other than a 10-count), there will be type of Insurance errors. In case A, the index will call for insurance when it is not warranted. In case B it misses insurance in some situation where it is called for.

If you are using the �Correct� index, then you will have both of types of errors. You have suggested that they �offset� each other. I think that the EV lost to Type A errors will be the same as the EV of type B errors, for the optimal index.

Let me think about for a while. I am not quite sure about the EV balancing. It has been a while since I thought about this, so I don�t want to say anything definitive until I can think about ti some more.

I think the answer to your question is YES. For any count (other than a 10-count), there will be type of Insurance errors. In case A, the index will call for insurance when it is not warranted. In case B it misses insurance in some situation where it is called for.

If you are using the �Correct� index, then you will have both of types of errors. You have suggested that they �offset� each other. I think that the EV lost to Type A errors will be the same as the EV of type B errors, for the optimal index.

Let me think about for a while. I am not quite sure about the EV balancing. It has been a while since I thought about this, so I don�t want to say anything definitive until I can think about ti some more.