When Does Bad Counting Become Worse Than No Counting?

[vonQuux]

Suppose someone is using BS and counting (HI-LO) but not using indices.

At what point does counting errors cause one to lose more than if one didn't bother counting at all? +/-1? +/-2? +/-5?

I'm asking because if I'm at a table and I lose count for some reason, I want to know how "uncertainty" I can rack up before I should either return to straight-up BS or leave the table.

I'm guessing that I should go with what I absolutely "know" and regard the "uncertains" as cards not seen. But suppose I know there were a couple of high cards, I just don't know how many?

TIA,
vQ

 

[godeem23]

Suppose someone is using BS and counting (HI-LO) but not using indices.

At what point does counting errors cause one to lose more than if one didn't bother counting at all? +/-1? +/-2? +/-5?

I'm asking because if I'm at a table and I lose count for some reason, I want to know how "uncertainty" I can rack up before I should either return to straight-up BS or leave the table.

I'm guessing that I should go with what I absolutely "know" and regard the "uncertains" as cards not seen. But suppose I know there were a couple of high cards, I just don't know how many?

TIA,
vQ

Since you're not using indices, you're only using your count to vary bet size. If that's the case, you'll be following exact BS and therefore win/lose/push the exact same percentage of hands as you would have if you weren't counting. The only difference will be the size of your bets. Since you're equally likely to overestimate the count as underestimate, your "bad counting" is no worse and no better than playing BS. If I misunderstood your question I'm sorry. Also, if any pros notice a flaw in my logic please straigten me out.

 

[Renzey]

Since you're not using indices, your win/lose/push percentage of hands will be the same as if you weren't counting. The only difference will be the size of your bets. Since you're equally likely to overestimate the count as underestimate, your "bad counting" is no worse and no better than playing Basic Strategy

I think his overall performance will be better than flat bet basic strategy because he'll raise his bets only when he thinks the count is positive -- never when he thinks it's negative. Thus, his betting errors will centralize around a positive count.

I see one caveat however. Since a given negative count produces a greater negative EV than the positive EV for an equally positive count (not to mention that a smaller negative count occurs more often than a larger positive count), ramping up at too slight an advantage could theoretically produce a net disadvantage after all errors blend together. Saving the ramp up starting point for about a +2.25 TC minimum (but jumping right to a 3 or 4 unit bet) will make it less likely that his mistake will have him betting into a minus EV.

 

[EasyRhino]

I think it depends on how well correlated your betting is to the count.

If your correlation is +1.0, it means you're sizing your bets perfectly. So your advantage would be the "usual" advantage through counting.

If your correlation is 0.0, then your bets are effectively random. Note that flat betting also counts as a correlation of zero. In this case, since your probably betting more than you were when you were flat betting, it would definitely increase the amount of money lost (it would be the amount of the increase in average bet).

If your correlation is -1.0, then you're betting exactly the OPPOSITE of the count, and have managed to magnify the house advantage magnificently.

So... it's the break-even point is probably somewhere in the positive, but low, category. Like 0.4 (40% correlation to the count), where the house edge becomes zero, but your variance has increased due to betting more.

 

[jimpenn]

If I lose count during end part of game and shoe is at least +2TC then I'll count down the small verses high cards after each remaining hand. If this continues to remain positive I'll spread 3/1 - 5/1. For example, if shoe has a true count of +2 with two decks remaining (8D) I'll play green verses red remainder of shoe, while making mental counts of each round of remaining cards.

 

[mdlbj]

He has lost the count, he is not going to know when to wipe the beer off his chin.

I think his overall performance will be better than flat bet basic strategy because he'll raise his bets only when he thinks the count is positive -- never when he thinks it's negative. Thus, his betting errors will centralize around a positive count.

I see one caveat however. Since a given negative count produces a greater negative EV than the positive EV for an equally positive count (not to mention that a smaller negative count occurs more often than a larger positive count), ramping up at too slight an advantage could theoretically produce a net disadvantage after all errors blend together. Saving the ramp up starting point for about a +2.25 TC minimum (but jumping right to a 3 or 4 unit bet) will make it less likely that his mistake will have him betting into a minus EV.

 

[Automatic Monkey]

Suppose someone is using BS and counting (HI-LO) but not using indices.

At what point does counting errors cause one to lose more than if one didn't bother counting at all? +/-1? +/-2? +/-5?

I'm asking because if I'm at a table and I lose count for some reason, I want to know how "uncertainty" I can rack up before I should either return to straight-up BS or leave the table.

I'm guessing that I should go with what I absolutely "know" and regard the "uncertains" as cards not seen. But suppose I know there were a couple of high cards, I just don't know how many?

TIA,
vQ

You have to make an impossible number of counting errors to not have an advantage in blackjack, well over 50% of the time. Missing 1-5 cards per 8 deck shoe causes your advantage to drop by an amount so small it might not ever become statistically significant in a lifetime of play. Assuming you are using High-Low, the average value of "cards not seen" is always zero, so adjust your count by exactly that- zero.

My advice is to not even think about making errors. It's like thinking about not falling off a bicycle, or trying not to swallow air when you eat. It only makes it worse. I know where you play and you've probably seen me playing there too- most important thing to do for success is just avoid playing bad counts.

 

[mdlbj]

Errors are ok. Come to my house and buy in, I will deal to you all night.

WOW....

 

[mdlbj]

How do you know if it is a bad count if you are making errors counting?

You are a ass.

Next time you try to quote SW, be more well, accurate.

You have to make an impossible number of counting errors to not have an advantage in blackjack, well over 50% of the time. Missing 1-5 cards per 8 deck shoe causes your advantage to drop by an amount so small it might not ever become statistically significant in a lifetime of play. Assuming you are using High-Low, the average value of "cards not seen" is always zero, so adjust your count by exactly that- zero.

My advice is to not even think about making errors. It's like thinking about not falling off a bicycle, or trying not to swallow air when you eat. It only makes it worse. I know where you play and you've probably seen me playing there too- most important thing to do for success is just avoid playing bad counts.

 

[Automatic Monkey]

How do you know if it is a bad count if you are making errors counting?

Because counting errors are the same as unseen cards. There is no difference. We can count and play with an advantage even though we will miss 25% of the cards in the shoe, every shoe and we can only hope that the cards in the next hand are a reasonable sample of the remainder of the shoe.

You are a ass.

Now that wasn't very nice. Is that the way the BJI guys taught you to speak to prospective customers?

 

[QFIT]

See http://www.blackjackincolor.com/blackjackerrors2.htm

 

[Bojack1]

See http://www.blackjackincolor.com/blackjackerrors2.htm

I like the sim it speaks volumes on how minimal mistakes in counting can hurt you. The problem I see is the graph shown is not indicitave of a player that is making a more realistic amount of mistakes for what would be the norm of most casual players. This sim represents a very good player, not one prone to mistakes. I would bet that very few players here playing in a casino make as little as 1 counting error per 100 rounds. Indeed the mistakes are usually much more, as well as extrapilated so that it is common for the casual, or mistake allowing counter to be off the count most times by much more than 1. Its true minimal mistakes will not truly hurt your game, thats why when we test our players we allow for a small amount of them. But I don't think from what I've seen out there, that this sim is a true representation of the majority of players.

 

[QFIT]

But I don't think from what I've seen out there, that this sim is a true representation of the majority of players.

Good point. But then, if someone is making more than five errors an hour, I submit that he/she is not a card counter.

 

[vonQuux]

See http://www.blackjackincolor.com/blackjackerrors2.htm

Thanks QFIT, I think that's what I'm looking for. But I'm not entirely sure I understand it.

If I'm reading that correctly, any counting errors will make my play worse than if I didn't count at all, which makes me think that I'm not quite understanding the underlying machinations of SCORE.

Can you elaborate a bit on this?

TIA,
vQ

 

[EasyRhino]

Naw, the basic SCORE figures on that chart, are the risk-adjusted expected earnings on theoretical $10k bankroll. Notice how they're all positive? It's all still in the expected win column.

So... I guess it would take more radical bet errors to wipe out an advantage than merely being off by +-1 a good chunk of the time. Like simply neglecting to notice there's a positive count, betting the minimum, and doing this about 50% of the time.

 

[vonQuux]

Naw, the basic SCORE figures on that chart, are the risk-adjusted expected earnings on theoretical $10k bankroll. Notice how they're all positive? It's all still in the expected win column.

So... I guess it would take more radical bet errors to wipe out an advantage than merely being off by +-1 a good chunk of the time. Like simply neglecting to notice there's a positive count, betting the minimum, and doing this about 50% of the time.

I hear ya but ...I'm trying to figure out the precise point where bad counting is worse than no counting.

In other words, just for argument's sake, suppose I'm at a $10 table with a $5,000 bankroll. Also suppose that I keep a perfect count, bet perfectly and play perfect BS over 1,000 hands. I should be able to calculate my expected profit (although my results will, of course, have variance).

Now suppose that everything else stays the same but I purposefully shift my count +1, then calculate my expected profit. Clearly that return will be less than my previous calculation. What if I try +2? +3? How about those numbers in the negative?

At what point does my shift cause my expected rate of return to be less than if I never bothered to count at all?

I'm trying to figure this out because once I know this number, I have a "grade" that I can shoot for. Suppose the answer to my question is "you can be between -2 to +3 on your count and -- all other things being equal -- you'll still have an expected return larger than no counting." Great. Now I know I shouldn't step into a casino until my counting gets within this margin of error.

As of this moment, I don't know what my allowable margin of error is. I have no real trouble counting accurately and at an acceptable speed (around 30 seconds per deck and improving daily). It's the "remembering-it-while-I'm-being-distracted part that's causing me errors and thus my query.

vQ

 

[rukus]

qfit's two posts are right on - if you make "minimal" mistakes it wont affect you very much. if youre making anything more than "minimal" mistakes, then you shouldnt be playing in a real casino yet anyway (but hey, it's your money to lose i guess).

 

[Automatic Monkey]

See http://www.blackjackincolor.com/blackjackerrors2.htm

I think those graphs prove my point. The most serious errors include Basic Strategy errors, which I'm assuming we're never making, but still a player can screw up Basic Strategy 5% of the time and be making half the money as if he didn't. With the other types of errors- how many millions of hands do you have to play before the difference in a game with SCORE of 40 and of 43 become statistically significant?

The kinds of errors I assume VonQuux are talking about are simple missed-a-card counting errors, which are the same as lost penetration. I usually sim these by treating them as unseen burn cards.

 

[EasyRhino]

Aim higher vQ. Over-engineer it. Build some margin into your game.

During practice type situations, where the environment is more controlled and you can check yourself, you really shouldn't be making mistakes in the count, or mistakes in play, of any sort.

Although, I have a suspicion that, once someone goes through the effort and learns hi-lo, that the more common ways to become a loser are overbetting, or too much play on non-advantage games. We need the Gallup people to do a poll.

The question of sheer counting errors would probably apply more to "half smart" players who are merely "watching cards", or only keep track for the first three hands or something.

 

[vonQuux]

I sincerely appreciate everyone's input and there's certainly some good advice contained therein. And so I don't mean to be obnoxious and I DEFINITELY don't want to seem like I'm demanding that someone do the math for me, but I do want to point out that the question remains unanswered.

That is, precisely how far off does my count have to become before the errors cause more harm than if I wasn't counting at all?

It looks like QFIT has provided an answer to a very similar question (how does improper counting affect SCORE) but due to my very poor math skills, I still don't know how this translates to a specific number for my specific set of betting conditions.

The reason I'm being a bit tenacious is because I'm learning that in blackjack, things are not always as obvious as they might seem. People are offering advice but if I don't know that number, I have no way of knowing the quality of that advice. Perhaps the number is extremely small, maybe there is a wide latitude for my count to be sub-optimal and still be better-than-nothing-at-all, I don't know.

If the deviation is large, that's good news for learners. If the number is quite small then it would probably save a lot of players a lot of money by knowing that they need to either be quite accurate or they're doing themselves more harm than good.

In either case, that break-point is quite useful. And while all the advice is helpful and appreciated, I'm still confused about where that point lies, exactly.

vQ