House edge = off the top edge?

Kasi

Well-Known Member
#21
FLASH1296 said:
Sometimes I am mistaken, but in this case, I am quite correct.

The Basic Strategist loses on average a computable percentage, not an amount, per hand played. The e.v. when computed includes the normal expected frequency of splits, double downs, and blackjacks. Please re-read this. It is essential to the nature of the game. Misunderstanding this basic fact leads to further misunderstandings re: blackjack.
It's not "per hand played", it's "per initial bet". In other words, per "round".

A flat-betting BS player will lose on avg the "HA*initial bet" no matter how many times he may split or double that initial bet.

Like you say, that HA takes into account all the doubling and splitting, etc.
It's just always expressed as a % of initial bet for that "round" no matter how many "hands" may result from that initial bet.

Understanding this basic fact can't hurt anything, that's for sure.
 

Kasi

Well-Known Member
#22
FLASH1296 said:
kasi,

You said: "Just curious, where in the world do you get these ideas from?

After counting all these years, does it actually, even subjectively, seem to you you spend an equal amount of time in neg TC's as pos TC's?"


Here is a simple explanation that will prove my point. Let's looks at the simplest balanced count in existence. It will be the same for the most complex count as well. I will use the Ace-Five Count. The 5's are tagged as +1 and the Ace's are tagged at -1. This is actually good enough to beat a (no longer existent) Single Deck S17 DAS game with a non-threatening bet spread. Interestingly, the running count can not be greater than +4 or less than -4. At the "top of the deck" and at any time there is a ZERO True Count (or Running Count) - The probability of a hand having an Ace is the same as the probability of the hand having a 5. Correct ? Therefore the count will move up and down with the depletion of Aces and Fives. By the bottom of the deck the count MUST return to ZERO [if all cards are dealt.] High and Low cards are equally present in the deck or the shoe and the + and - True Counts MUST exist equally. The reason that the Basic Strategist has a losing situation is that he starts off with a disadvantage that we call the "House Advantage". Therefore, the basic strategist will only be "getting the best of it" when the True Count is +1 or +2 (Hi-Lo) depending on the size of the house edge.

You statement about "... subjectively ..." is accountable by what psychologists call "selective memory" If you are a poker player you will recall the bad beats you have suffered but will not recall many of the bad beats that you have been on the winning side of. Ask any ploppy if he or the dealer gets more blackjacks and 9 out of 10 will look at you like you are crazy, as they say, "Of course the dealer always gets more BJ's than I do." OBVIOUSLY, the dealer, you, me, and everyone else receive one "snapper" in every 21 hands. Think about it.


Eshew Subjectivity. Embrace Objectivity.


"...there is a method to scientific thinking and it includes being constantly vigilant against self-deception and being careful not to rely upon insight or intuition in place of rigorous and precise empirical testing of theoretical and causal claims."
Well, I'm thinking about it lol.

How do you explain every single table in Don's BJAIII having a greater frequency of -TC's than +TC's?

Think about that any counting system is meant to reveal when an advantage occurs. The game is neg to begin with. If the count system is doing it's job, doesn't it seem reasonable it will spend more time identifying neg situations than pos?

Obviously alot of one's equity comes from the 10% of the time when one doubles or the 1/21 times one gets a 3-2 BJ payoff. They don't happen that often but when they do, the pay-off helps avg things.

Think about that maybe lol.

Or, run a billion round sim with a balanced count system, and you tell me if if a -1TC occurs as frequently as a +1TC -2 vs +2 etc.

Use one of those free ones if you want since, I assume, even after 40 years of counting, you still haven't blown $100 on a sim.

I dion't know why it's always me that seems like the bad guy lol.

Apparently, it's only you and me that have an opinion on what HA means, whether TC's of -1 occur as frequently as TC's of +!, etc :)
 

1357111317

Well-Known Member
#23
Kasi I'm glad to see that you agree with me on the HA part of the discussion however I think Flash is right when it comes to the TC's. There is an equal chance that the TC would be +1 or -1, except at my table where the count is always negative. But at all non shuffle tracked tables the Odds of having a +1 TC or -1 TC should be the exact same. The advantage will always be in the houses favour way more often obviously but in terms of TC's there is an equal chance of a negative and an equally positive TC occuring.
 

Kasi

Well-Known Member
#24
1357111317 said:
Kasi I'm glad to see that you agree with me on the HA part of the discussion however I think Flash is right when it comes to the TC's. There is an equal chance that the TC would be +1 or -1, except at my table where the count is always negative. But at all non shuffle tracked tables the Odds of having a +1 TC or -1 TC should be the exact same. The advantage will always be in the houses favour way more often obviously but in terms of TC's there is an equal chance of a negative and an equally positive TC occuring.

That's cool lol.

I agree with you 100% on the HA stuff and 0% on the the other stuff lol.

On the other stuff, here's a 1 billion sim from Psim using Hi-Lo:

TC Freq
-20 1904
-19 3013
-18 6528
-17 14205
-16 33225
-15 65899
-14 120879
-13 262527
-12 508996
-11 886309
-10 1629608
-9 2717978
-8 4995524
-7 8286363
-6 13990214
-5 22992027
-4 40589843
-3 68381683
-2 120764496
-1 182275959
0 257926684
1 114472031
2 66018256
3 38037628
4 22919080
5 13291625
6 8078254
7 4609886
8 2767824
9 1533211
10 861472
11 458225
12 243531
13 137453
14 63051
15 28454
16 15182
17 6145
18 2915
19 1120
20 793

1,000,000,000

As anyone can see, a TC-1 occurred 182MM times compared to a TC+1 114MM times.

A TC-2 occurred 121MM times vs a TC+2 of 66MM times.

A TC-5 occurred 23MM times vs a TC+5 occurring 13MM times. Etc.

A -TC occurred 469MM times vs a +TC of 273MM times.

This was for a 6D shufflepoint 249 using Hi-Lo.

Not sure what else I can do to convince you and Flash on this point?

But, if either of you have something to convince me otherwise, I'm all ears.

I'd prefer maybe something more than that there are the same number of Aces in a deck as there are 5's, as excellent as that point may be.
 

Thunder

Well-Known Member
#25
Logically, how does that make sense for there to be more - counts than positive counts if it's a balanced count? I see the data clearly says there are more negative counts than + ones but it just doesn't make sense. Perhaps your simulator is off?
 

sagefr0g

Well-Known Member
#26
Thunder said:
Logically, how does that make sense for there to be more - counts than positive counts if it's a balanced count? I see the data clearly says there are more negative counts than + ones but it just doesn't make sense. Perhaps your simulator is off?
here's some true count frequencies from Wong's Professional Blackjack.
http://www.blackjackinfo.com/bb/showpost.php?p=117223&postcount=3
note: i fudged the tc=0 numbers, Wong had tc=-0, tc=0 & tc=+0 frequencies as well, i just combined them in the sheet.

kind of guessing here, but for a simulation thats got players against a dealer and there is a cut card involved, then the 'cut card effect' may be involved as far as having slightly more negative tc's than positive.
http://www.blackjackincolor.com/blackjackeffects1.htm

also, maybe how you mathematically figure the true count, rounding down may skew it a bit?:confused:
 

1357111317

Well-Known Member
#27
Kasi said:
That's cool lol.

I agree with you 100% on the HA stuff and 0% on the the other stuff lol.

On the other stuff, here's a 1 billion sim from Psim using Hi-Lo:

TC Freq
-20 1904
-19 3013
-18 6528
-17 14205
-16 33225
-15 65899
-14 120879
-13 262527
-12 508996
-11 886309
-10 1629608
-9 2717978
-8 4995524
-7 8286363
-6 13990214
-5 22992027
-4 40589843
-3 68381683
-2 120764496
-1 182275959
0 257926684
1 114472031
2 66018256
3 38037628
4 22919080
5 13291625
6 8078254
7 4609886
8 2767824
9 1533211
10 861472
11 458225
12 243531
13 137453
14 63051
15 28454
16 15182
17 6145
18 2915
19 1120
20 793

1,000,000,000

As anyone can see, a TC-1 occurred 182MM times compared to a TC+1 114MM times.

A TC-2 occurred 121MM times vs a TC+2 of 66MM times.

A TC-5 occurred 23MM times vs a TC+5 occurring 13MM times. Etc.

A -TC occurred 469MM times vs a +TC of 273MM times.

This was for a 6D shufflepoint 249 using Hi-Lo.

Not sure what else I can do to convince you and Flash on this point?

But, if either of you have something to convince me otherwise, I'm all ears.

I'd prefer maybe something more than that there are the same number of Aces in a deck as there are 5's, as excellent as that point may be.
Well I guess the sim doesnt lie. I just don't see why more 10's and aces are coming out of the shoe than 2-6's. It just seems like there should be an equal chance of each of them coming out of the shoe. anyone have anything that could explain this?
 
#28
1357111317 said:
Well I guess the sim doesnt lie. I just don't see why more 10's and aces are coming out of the shoe than 2-6's. It just seems like there should be an equal chance of each of them coming out of the shoe. anyone have anything that could explain this?
Hmmm...I have a theory - may be 1,000,000 are not enough.
May be the next 1,000,000, the all negative counts and all positive counts will swich ther roles, and the positive ones will become more often.Than take the results of the first billion and the second one, the positives and negatives will be almost even, as it must be ! :cool:
 

FLASH1296

Well-Known Member
#29
Quelle Bizarre!

These results cannot occur with a legit shoe

A short sim can and will lie.
1,000,1000,000 is not only too short,
but the results are so absurdly
"off" that I question both the reliability and
the validity of the results.

Indeed, if the results cited were exemplary
of real life, ordinary card counting would be
a monumental exercise in futility.
 

Sonny

Well-Known Member
#30
Kasi said:
So, even spreading to 4 hands in a round, doubling each one, and having $80 out here for that round (don't call it 4 hands), you still expect to lose on average for that round 6.6 cents.
FLASH1296 said:
NO. Spreading to four $10 hands will cost the basic strategy player four times 6.6 cents or 26.4 cents;
precisely the same as if there were $40 wagered on one hand.
I think the confusion here is over how you arrived at those four hands. If you spread to 4 hands off the top, before seeing any of your cards, then Flash is right, your EV is equal to the house edge on each hand. However, if you played one hand and then split it three times then the EV is much different because you have already seen your cards and are acting based on that information. The fact that the player doubled on all of them means that the EV was actually much higher than normal, but a BS player would not know that until after they saw their cards. It is the difference between absolute EV and conditional EV. I think the two of you are talking about different situation but you are both right.

As for the higher number of negative counts compared to positive counts, it all depends on how you calculate your TC. If you are flooring, as many people do, then you will be brining everything from +0.001 to +0.99999 down to 0. The number of +1 TCs will be lower because a lot of those positive counts fell into the zero category. Similarly, everything from –0.01 to –0.999 will be counted as –1 so that will add to the number of negative counts.

Try running some sims with different TC calculation methods (especially rounding) and see how they compare.

-Sonny-
 

k_c

Well-Known Member
#31
Kasi said:
That's cool lol.

I agree with you 100% on the HA stuff and 0% on the the other stuff lol.

On the other stuff, here's a 1 billion sim from Psim using Hi-Lo:

TC Freq
-20 1904
-19 3013
-18 6528
-17 14205
-16 33225
-15 65899
-14 120879
-13 262527
-12 508996
-11 886309
-10 1629608
-9 2717978
-8 4995524
-7 8286363
-6 13990214
-5 22992027
-4 40589843
-3 68381683
-2 120764496
-1 182275959
0 257926684
1 114472031
2 66018256
3 38037628
4 22919080
5 13291625
6 8078254
7 4609886
8 2767824
9 1533211
10 861472
11 458225
12 243531
13 137453
14 63051
15 28454
16 15182
17 6145
18 2915
19 1120
20 793

1,000,000,000

As anyone can see, a TC-1 occurred 182MM times compared to a TC+1 114MM times.

A TC-2 occurred 121MM times vs a TC+2 of 66MM times.

A TC-5 occurred 23MM times vs a TC+5 occurring 13MM times. Etc.

A -TC occurred 469MM times vs a +TC of 273MM times.

This was for a 6D shufflepoint 249 using Hi-Lo.

Not sure what else I can do to convince you and Flash on this point?

But, if either of you have something to convince me otherwise, I'm all ears.

I'd prefer maybe something more than that there are the same number of Aces in a deck as there are 5's, as excellent as that point may be.
In the weekend warrior thread my sims seemed to indicate a tendency toward a greater frequency of negative counts. I was wondering how that could be and think possibly it could be related to the use of a cut cut. It would seem to me that if all cards are dealt then the tendency would be an equal number of positive and negative running counts. That being said, it appears to me that the sim you posted gives a larger ratio of neg/pos than I would expect, but I could be wrong.
For 13018 rounds ave hilo TC=-.23. You might expect it to be 0.
For 13018 rounds ave TKO TC=-4.28. You might expect it to be -4.
 

Attachments

QFIT

Well-Known Member
#32
As Sonny said, the TC frequencies depend on the method of TC calculation. There are many, many methods of TC calculation including lots of combinations of those described at http://www.qfit.com/CalculatingTrueCounts.htm and all of these change the frequencies. Penetration, cut card and number of players also affect TC frequencies. The third chart at http://www.blackjackincolor.com/truecount4.htm shows the differences between Truncate, Floor and Round. I also have three charts at http://www.blackjackincolor.com/truecount1.htm.

All the tables in BJA Chapter 10 use Flooring with one-half deck estimation for shoes and one-qtr deck estimation for pitch games.
 
Last edited:

Kasi

Well-Known Member
#33
Sonny said:
I think the confusion here is over how you arrived at those four hands. If you spread to 4 hands off the top, before seeing any of your cards, then Flash is right, your EV is equal to the house edge on each hand. However, if you played one hand and then split it three times then the EV is much different because you have already seen your cards and are acting based on that information. The fact that the player doubled on all of them means that the EV was actually much higher than normal, but a BS player would not know that until after they saw their cards. It is the difference between absolute EV and conditional EV. I think the two of you are talking about different situation but you are both right.

As for the higher number of negative counts compared to positive counts, it all depends on how you calculate your TC. If you are flooring, as many people do, then you will be brining everything from +0.001 to +0.99999 down to 0. The number of +1 TCs will be lower because a lot of those positive counts fell into the zero category. Similarly, everything from –0.01 to –0.999 will be counted as –1 so that will add to the number of negative counts.

Try running some sims with different TC calculation methods (especially rounding) and see how they compare.

-Sonny-
Thanks Sonny - I'm down with that spreading to 4 separate hands off the top lol.

I just got thrown off a little, maybe, when Flash called 100 rounds a $1000 "handle". I assumed by "handle" he meant total dollars actually wagered after 100 rounds so got caught up in splitting to 4 "hands" in one round lol.

As for TC's equaling each other, I guess I've just never seen it happen it sims lol. But it does sound logical they could be lol. I looked at Norm's charts the other nite and they didn't look as symetric to me then as they do now lmao.

Maybe it's like some said, the cut-card screws up theory vs real life lol. Or maybe it's like k_c suggested that one would have to have 100% pen to make it happen?

Maybe, like Norm siad, even the number of players might make it impossible for that to happen?

I guess in other words maybe just a question for Norm as to what he's have to tell his CVCX or CVDATA to have a sime come out with TC-1=TC+1, etc in freq?

I don't have software that can round TC's lol.

What would be necessary for -Tc's to = +TC's in frequency even in theory maybe lol?

Maybe a counting BS player playing a fixed number of rounds per shoe that would ensure never reaching a cut-card heads-up?

I think Psim does a TC to the exact card as impractical as that may be.
 

QFIT

Well-Known Member
#34
Oddly

Kasi said:
I guess in other words maybe just a question for Norm as to what he's have to tell his CVCX or CVDATA to have a sime come out with TC-1=TC+1, etc in freq?
Actually, there is no case where this is true. If you truncate and deal a fixed number of rounds, you will start more rounds with a -1 TC than a +1 TC. But, you will start fewer rounds with a -10 TC than a +10 TC. Sounds contradictory. But there is a reason.

More rounds end with a ten than any other card causing an increase in -1 TCs. But, rounds that end with small cards often have many small cards so large positive TCs are more common than large negative TCs.
 
#35
QFIT said:
More rounds end with a ten than any other card causing an increase in -1 TCs. But, rounds that end with small cards often have many small cards so large positive TCs are more common than large negative TCs.

Why would that give you more -1 TCs? Sure, there are more 10s in the deck, but there are an equal number of 2-6s compared to 10s and Aces. So they should cancel each other out and you end up with an expected count of 0.

Also, why would the cut card matter for this situation? Whether you see 1 out of 6 decks or all 6 decks, the count should still hover around 0. I thought that's why you can join a shoe game several decks in and start counting from 0. The deeper the cut card, the better the information, but I just don't see how that would change the number of +/-1 TCs you see in a shoe.

Sonny's answer is the only one that makes sense to me. Unless I'm missing something here.
 

QFIT

Well-Known Member
#36
The average true count is zero. However, there are more true counts of -1 than +1 because more often a round is ended when a ten is dealt. This is not because there are more tens. It is because a ten is more likely to bust or complete a hand. The cards, on average, are evenly distributed. But the round stopping rules make it more likely to end on a down note. A deuce is quite unlikely to end a round.

The cut-card matters because of the cut-card effect. There are more rounds that begin with a negative count when you have a cut-card. See http://www.blackjackincolor.com/blackjackeffects1.htm.

Below is a chart of the percentage of counts from -7 to +7. Red is positive counts and green is negative counts. For counts of 1 and 2, there are more negatives than positives. For counts of 3 and higher, there are more positive counts.

 
Top