Acceptable error rate

[NotEnoughHeat]

While imagine different people have different ideas on what an acceptable, standard, and average error rate is like (think about the two big names associated w/ the drama on the site and refer to 'approaches' etc.), I nevertheless would like to hear thoughts on this.

I'm not sure if there's any industry standards for measurements of error, but in any case I would imagine it being X errors per Y hands (or any other measure preferably round based instead of time).

I imagine no one expects themselves or anyone else to be a perfect counting robot that only makes a low single digit number of mistakes per millions of rounds, but what's everyone's idea of sufficient, average, good/ideal errors per 100 rounds (or whatever you want to use) for:
A) Counting
B) Betting Strategy & Playing Strategy
C) TC conversion (which might have issues quantifying due to desk estimation)
D) Combinations of the above

Is on average, 1 counting mistake per 6D shoe acceptable? is 1 for every 4 sufficient?

Even if you don't bother to measure your own error rate, just blurt out what you think.

Hopefully there can be some good discussion about the varying answers and perhaps the question itself.

[ZenMaster_Flash]

Errors are simply NOT acceptable.

Our edge is too small to accept errors.

[Freightman]

Items A B & C are each seperate issues.
The drama referred referred to is in regards to running count. KJ is adamant the human mind can only absorb -1,0,+1 on a long session without error or fatigue. Tthree claims he can keep track of 2 linear counts which combine both PE and BC. I personally have no problems whatsoever keeping track of -1,-.5,0,+.5,+1.0,+1.5. I believe Tthree keeps dual counts, but would suggest, on a heads up basis, he would need to some degree, control the speed of the table. He can confirm or adjust this point. I can absolutely guarantee that I can easily maintain perfect count to exceptional dealer speed. The fastest heads up shoe I've ever had is 1-6d shoe dealt to 1 deck in 6 minutes. I would happily wager KJ to an accuracy speed contest. He would likely lose.

Betting and playing strategy. Tarzan crunches so much info, it slows him down. Just yesterday, he indicated a 35 second single deck count. That's slow, but think about what he crunches. Betting strategy is subject to judgement, which I believe Tarzan possesses to the nth degree, which I believe, though may be wrong, that KJ possesses little of, and myself, for sake of argument is somewhere in between. What I'm saying here is that 0 judgement and level one count gets the money, but subject to pretty heavy duty swings. By judgement, I'm mostly, but not completely referring to how one handles intermediates. I'm also heavily referring to not being a blind slave to the true count (complicated and another thread)

TC conversion - a couple of thoughts. I quite often don't look at the discard tray - I know by rounds played and running count what the approx TC (good and interesting arguments fir the one who decides to challenge this) I will definitely refer to the tray in high bet situations and for significant index plays. Secondly, keep in mind that different stores use different thicknesses of cards. You need to adjust your eyeball measurèments to account for this.

In summary, I thing the big argument refers to running count - do - be creative, don't be robotic.

[Freightman]

Errors are simply NOT acceptable.

Our edge is too small to accept errors.

I do - I make some on purpose - always on single unit bets. Let me run a scenario here.

Your being hawked - your bankroll is absolutely massive for the game being played. You make silly errors at shit counts. Tou have 10 units at true 4 or 5. You have a 15 unit bet as the count is coming down, I think - in fact I know that you catch the drift.

[AndretheGiant]

Oh please Flash ,are you saying you have never made a mistake?

[Stealth]

Errors are both NOT ACCEPTABLE and inevitable.

If you are a lone wolf, then perhaps you perform with a very low error rate, but I can assure you it is not zero. Maybe it is a simple as missing the TC conversion or not realizing you moved to the table with surrender, whatever, they happen.

In a team environment, were they are moving on the strip from casino to casino, game to game, it is not hard to miss some key variable. My experience was that they range from not being aware of the specific rules for a game, missing a surrender play, thought the pen was better than the game I sat at, to being called in by a back counter and going to the wrong table to name a few.

In my research and experience, when long term (>n0) team results are reviewed, often actual results were some 60-80% of EV. Although I have no empirical evidence to support such a statement, antidotal evidence does. Common sense tells us that we can not, and do not, play with the efficacy of a computer running a 3 billion hand sim, in fact, our actual results (hands played correctly, not win/loss) must be worst than the sim by the very nature of the human versus computer comparison.

Since we have no way (other than casino verite and then only in a test environment) to validate our correctness in actual play then we can only postulate what those error rates might be.

I am very interested in the perspectives on this subject and will tell you that I believe it to be a reduction in EV by a range of 20-30%.

Now, I expect the purist among us to "blow me up" with their absoluteness of their quality of play, I will just ask, how do you know?

Let the games begin!

[ZenMaster_Flash]

" ... are you saying you have never made a mistake?"

What I said NEITHER indicates nor implies such.

[ZenMaster_Flash]

"I do - I make some on purpose ..."

An error is a mistake.
Cover Plays are ...
anything but mistaken.

[Bigdaddy]

Errors are both NOT ACCEPTABLE and inevitable.

If you are a lone wolf, then perhaps you perform with a very low error rate, but I can assure you it is not zero. Maybe it is a simple as missing the TC conversion or not realizing you moved to the table with surrender, whatever, they happen.

In a team environment, were they are moving on the strip from casino to casino, game to game, it is not hard to miss some key variable. My experience was that they range from not being aware of the specific rules for a game, missing a surrender play, thought the pen was better than the game I sat at, to being called in by a back counter and going to the wrong table to name a few.

In my research and experience, when long term (>n0) team results are reviewed, often actual results were some 60-80% of EV. Although I have no empirical evidence to support such a statement, antidotal evidence does. Common sense tells us that we can not, and do not, play with the efficacy of a computer running a 3 billion hand sim, in fact, our actual results (hands played correctly, not win/loss) must be worst than the sim by the very nature of the human versus computer comparison.

Since we have no way (other than casino verite and then only in a test environment) to validate our correctness in actual play then we can only postulate what those error rates might be.

I am very interested in the perspectives on this subject and will tell you that I believe it to be a reduction in EV by a range of 20-30%.

Now, I expect the purist among us to "blow me up" with their absoluteness of their quality of play, I will just ask, how do you know?

Let the games begin!

Thanks Stealth - a very reasonable and level-headed observation, based on your extensive experience. Who among us would not haircut their simmed EV by 20% to reflect reality?

[Symbiant]

I am a beginner and have about 50 hours of live play so far. Anecdotal data:

A, Counting: My home casino does not shuffle and reset shoes after players depart, so if I want to play, I usually have to come in to a partially-dealt one. When there are n cards already in the discard tray, it is exactly the same as forgetting to count the first n cards. This makes it harder to get a high TC, but with high casino tolerance I've found that it doesn't matter. I would guess that systemic RC errors would be problematic, but very occasional (and offsetting) ones would have a small impact on a shoe game. I find myself having to approximate the count on a big layout about once every 200 rounds right now, usually because I get distracted. I expect that number to go down a fair bit as I get better, especially at anticipating and reducing the number of distractions.

B, Betting: I don't use a rigid ramp, so "errors" are masked by natural cover anyway. This makes it harder to calculate EV ahead of time, of course.

B, Playing: I catch myself making a mistake about every two hours, usually towards the end of a longer session. I generally correct my mistake in time. I would guess I make less than one actual BS playing mistake every ten hours or 1000 rounds. I use only about 10 indices today, and I think I will miss the indices more often. BS is muscle memory at this point, so the errors are coming from visual issues (for example, yesterday I misread a 44 as an A4 for a moment).

C, TC conversion: I think this is inherently a little sloppy but self-balancing, unless you are a systemic mis-estimator in one direction. If you are using a more dynamic betting strategy anyway, I don't think it matters too much.


I agree with Stealth's suggestion to sandbag simulation EV by 20%. The problem is that deviation is so large, you'll need a ton of time in before you have any clue really.

[MJGolf]

I think Stealth nailed it on the head. From a lone wolf side, just HOW would you be able to truly gauge mistakes UNLESS you had the film from the surveillance camera's for the tables you played. That may be facetious but true. Or unless you had a "stalker" teammate watching and logging your every play.

I think a more accurate gauge or estimate of your play would be your own logging of hours, win rate and EV calculations. Just HOW far do you deviate from your expectations in the "long run"? If it's close, I bet your error rate is low. If it's too much different, then something has to be wrong and your play re-evaluated. We are ALL human.........while Flash states errors are "unacceptable" they are "inevitable".

[CHAM]

I also agree that Stealth nailed it. No way can I expect the results that I run on CVCX.

[NotEnoughHeat]

Just to re-frame the question a bit, sure no errors are 'acceptable', but you would think that it's not simply a yes-or-no situation and we might have different reactions to different levels of error (ideal, acceptable, I-can-live-wish-this, could-be-better, 'I should quit playing till I improve').

To follow that, a bit more on my mind: team recruitment standards. Obviously if you made someone do a ridiculous endurance test of counting 60 6D shoes in a row and the only error the guy made was fudging up the RC one time out of all 60 shoes, you wouldn't turn to him and say "Sorry, no errors are acceptable. Thanks for coming out". That's more of what I had in mind. Where do you draw the line? Perhaps you take his error rate in conjunction with other factors in determining his eligibility.