The Halves Count Challenge

[tthree]

I agree hi op II with A side count beats halves; it better, nearly twice the work for 5% to 10% more.

With Ao2 & halves what indices were used? I guess all?, which is why Ao2 closed the gap.

The sim details are pretty extensive but doesn't specify what indices were used. They do say they were generated from the SBA results file if that is useful information toward answering your question.

 

[tthree]

Assuming these simulations are accurate, why is Hi-Opt II suddenly outperforming AO2 and Halves? The ace side-count must raise the advantage insignificantly on a 6D game, so where are these changes coming from? And I thought Hi-Opt II was mainly a pitch game count.

I trust the accuracy of Qfit's data, but I also respect other simulations made as well due to both sims being ran by AP's. I'm really trying to step it up from TKO to a level 2 count and I was going to take FLASH's advice and work with Zen for a gain in advantage but I see contradiction through different oppinions and these simulations, so I'm really not sure which path I should take.

Can someone please explain?? Thanks.

HIOPT II uses the ace side count to adjust the RC for the betting TC conversion. It raises BC from .91 to a little over .98. BC is tremendously important in shoe games, need I explain further?

 

[tthree]

AO2 without Aces would perform no better than HO2 without Aces.
Neither would be competitive with Halves, without Aces. Both are obsolete relics. zg

Every comparison I have seen puts HIOPT II with ace side count at the top in all categories. I am not sure how it becomes a relic except to those who are easily swayed by propaganda. There is a lot of things to consider for what count best suits you as an individual. HIOPT II with ace side is not the best fit for many but for the people who it fits well it can't be beat. I don't understand how either can be considered a relic, especially not HIOPT II. With increasingly bad games being the norm that increase in your edge becomes more and more significant. Hopefully I don't need to spell that out with math, it should be obvious to everyone.

 

[blackjack avenger]

kids fighting in the sand box

Assuming these simulations are accurate, why is Hi-Opt II suddenly outperforming AO2 and Halves? The ace side-count must raise the advantage insignificantly on a 6D game, so where are these changes coming from? And I thought Hi-Opt II was mainly a pitch game count.

I trust the accuracy of Qfit's data, but I also respect other simulations made as well due to both sims being ran by AP's. I'm really trying to step it up from TKO to a level 2 count and I was going to take FLASH's advice and work with Zen for a gain in advantage but I see contradiction through different oppinions and these simulations, so I'm really not sure which path I should take.

Can someone please explain?? Thanks.

The A side count in hi op 2 is for betting, without it, its weak. The main count is high PE, good for DD. However u have to have the side count.
:cool2:

Probably the count u are using is fine. If want to go up a level, look for something similar to ur current count.

There is some differences in counts and specific games, but I am trying to show the dominance of halves among counts across games, not using a side count.

 

[Automatic Monkey]

See if you can find a count; without side count, that outperforms halves using SCORE and N0....

Kind of a useless challenge, but- try true-counted BRH-1 in a S17, NS game and true-counted Uston SS in a H17 game.

 

[Friendo]

QFIT sim results

I have recently run a few CVCX simulations comparing halves and Mentor, and have seen halves consistently behind mentor by a whisker. Both were used with plenty of indices under various conditions on 6-deck shoes.

I suspect this may be operator error. I don't use the standard Mentor indices, but I doubt it's my putatively-improved indices which are putting Mentor ahead. It could be an artifact of the 2-deck true count conversion: a fairer comparison would be to double the halves tags and use the same 2-deck true count method for both.

Due to the number of simulations I have seen which show halves beating Zen by a whisker in shoe games (e.g., in BJA3), and my general belief that Mentor and Zen are nearly identical in performance, I still believe that halves is a stronger system for shoes, but my simulation results show how close this whole question is.

It is definitely worth taking another look at this. I'll pore over all the CVCX parameters and re-run the comparisons with a 2-deck true count for both.

I also don't buy into the "level 3 counts are all ultra-hard" argument. Halves only has one tag making it a level 3 count, and that extra weight on the 5 is balanced by the ace being the same weight as a ten, an advantage not enjoyed by Zen and Mentor: with halves, 9 is the only oddball among the negative-weighted cards.

I probably won't switch, but I would be quite happy with halves.

 

[blackjack avenger]

not true to halves

I have recently run a few CVCX simulations comparing halves and Mentor, and have seen halves consistently behind mentor by a whisker. Both were used with plenty of indices under various conditions on 6-deck shoes.

I suspect this may be operator error. I don't use the standard Mentor indices, but I doubt it's my putatively-improved indices which are putting Mentor ahead. It could be an artifact of the 2-deck true count conversion: a fairer comparison would be to double the halves tags and use the same 2-deck true count method for both.

Due to the number of simulations I have seen which show halves beating Zen by a whisker in shoe games (e.g., in BJA3), and my general belief that Mentor and Zen are nearly identical in performance, I still believe that halves is a stronger system for shoes, but my simulation results show how close this whole question is.

It is definitely worth taking another look at this. I'll pore over all the CVCX parameters and re-run the comparisons with a 2-deck true count for both.

I also don't buy into the "level 3 counts are all ultra-hard" argument. Halves only has one tag making it a level 3 count, and that extra weight on the 5 is balanced by the ace being the same weight as a ten, an advantage not enjoyed by Zen and Mentor: with halves, 9 is the only oddball among the negative-weighted cards.

I probably won't switch, but I would be quite happy with halves.

Halves should be TC by deck

Probably mentor closes on halves due to many indices, as you state. Also, the 2 deck divisor hurts halves.

Little reason for most to switch counts.

To three:
Hi op 2 is considered outdated because of the lack of good hand held games which blunt the PE of the main count, which makes the A side count less meaningful. Also, the trend is to simplify.

 

[blackjack avenger]

in very good games

Halves with no side count outperforms hi op 2 with deeper penetration
In modern BJ chapter 8

 

[blackjack avenger]

help me out here

Yes, the goofball math geeks have hijacked this thread and continued their ongoing meaningless debate that is now consuming several threads in this section. Certainly wasn't my intent when I initiated the thread, and I am sure not Ken's when he 'stickied' it. :sad:

However the list of very experienced players, (sucker, 21forme, Ken Smith, Richard Munchkin, Qfit, ZenG) that have reached the same conclution is pretty impressive to me. These are players with a great deal of actual real play experience, some who play excusively for a living. There are several other's like Big Player, that while they haven't weighen in on this thread, have expressed similiar findings in other threads and discussions. Hopefully that will be an emphasis for newer players reading the thread.

So if someone makes a general statement that simplicity rules no one can challenge it?
One thing to realize is creating and selling counts is a business and the pendulum had swung pretty far toward complexity. Then it was realized simpler counts are effective. So has come the avalanche of simpler counts. Nothing wrong with simpler counts except

 

[blackjack avenger]

The Case For Higher Level Counts.

Info from cvvx online viewer.

6h17, das 4.5/6
Ill 18 indicies for hi lo and halves
13.5% ror
wonging
Optimized ramp for each count.

If one is going to make a claim for simplicity then perhaps a simple count should be considered.

Halves, CE $18.39, $10,000 bank
Hi lo, CE $18.43, $10,600 bank
KO rookie, CE $19.39, $20,000 bank

If one looks at bank vs CE simplicity seems to not measure up.
The difference in bank sizes for hi lo to beat halves is significant, about 30 hrs of play.

So one could use a higher level count and have a lower ror given a certain CE.

Below is subjective:
I am not an advocate of side counts.
Most should probably stick with the count they are using, complexity alone should not be a goal.

 

[tthree]

You don't have to convince me. Every case for simplicity is framed by unnecessarily handicapping a better system. The comparison breaks down on an even playing field so that we are back to complexity trumping simplicity. When I upgraded it was from HIOPT I to HIOPT II. I have always kept a side count. I can keep up with the dealers and playing conditions so I don't slow the game unless I want to. I get bored with simplicity and make more mistakes. I am sure I am an odd duck in that regard but that is me. The first 2 years at the University was a review of what I learned in High School. I didn't perform up to my standards because it was boring. Only when I was challenged did I shine.

 

[FLASH1296]

tthree's well-crafted post mirrors my own thoughts.

"Every case for simplicity is framed by unnecessarily handicapping a better system."

There is NO doubt that Level One Counts cannot compete with Level Two counts.

 

[zengrifter]

Every comparison I have seen puts HIOPT II with ace side count at the top in all categories. I am not sure how it becomes a relic except to those who are easily swayed by propaganda. There is a lot of things to consider for what count best suits you as an individual. HIOPT II with ace side is not the best fit for many but for the people who it fits well it can't be beat.

Most of the people it "fits well with" are deluded by the sims and their own subjective experience. Using the 1/4D ace-density estimation approach may look strong in sims but in practice will not yield better than ZEN or Mentor without the ASC. Thus working harder for the same net result.

In order to achieve the superior sim-indicated results you must use a secondary RC overlay: 2/5 +1 vs A -2 added to the primary count prior to TC adjustment, for betting (which is not practical for most users). zg

 

[iCountNTrack]

I have cleaned the thread and merged it with another one. Please make our jobs easier

 

[Sharky]

...The first 2 years at the University was a review of what I learned in High School...

:laugh:

 

[kewljason]

Info from cvvx online viewer.

6h17, das 4.5/6
Ill 18 indicies for hi lo and halves
13.5% ror
wonging
Optimized ramp for each count.

If one is going to make a claim for simplicity then perhaps a simple count should be considered.

Halves, CE $18.39, $10,000 bank
Hi lo, CE $18.43, $10,600 bank
KO rookie, CE $19.39, $20,000 bank

If one looks at bank vs CE simplicity seems to not measure up.
The difference in bank sizes for hi lo to beat halves is significant, about 30 hrs of play.

So one could use a higher level count and have a lower ror given a certain CE.

Below is subjective:
I am not an advocate of side counts.
Most should probably stick with the count they are using, complexity alone should not be a goal.

You guys want to keep going back to your computer generated simulations as the definitive answer to this equation. These results do not always translate into the real world of blackjack play as I have said on numerous occasions. You need to think outside the box a little.

Lets go back to an example I used earlier, that was first brought up by 21forme. Counting 2 sometimes 3 tables at once. I almost always am tracking at least one other table besides the game I am currently playing. If the count on the other table heads north I am on that table in a nanosecond. This basically gives me double the high count percentages, which your simulations don't show. This probably boost my overall win rate 50 to as much as 100 percent. So while your complex counts may boost your earnings 5-10% as simulations show, you probably aren't able to track several tables at once using a level 3 count like halves or some count with a side count or several side counts. And this is just one example of some things you are free to do by keeping things simple that you may not be able to do with a more complex count. In the real world, everything doesn't fit so nicely into a computer simulation.

As I have said on previous occasions, the discussion isn't about the math. The discussion is about whether it is worth the effort and if there are better ways to go. If you take the discussion outside the computer and into the real world, you will find a different answer. And as I said in another thread, which has since been removed, I find it very interesting that many long time, very experienced players, several who play exclusively for a living have come to the same conclusion. (I am not going to list names again.)

 

[tthree]

:laugh:

Yes I was a gifted student that learned way ahead of all the other gifted students. I set my own class schedule in high school. I learned over 2 years of college physics and maybe 3 years of advanced college math classes.

 

[tthree]

As I have said on previous occasions, the discussion isn't about the math. The discussion is about whether it is worth the effort and if there are better ways to go. If you take the discussion outside the computer and into the real world, you will find a different answer.

This will probably go over your head but I will try anyway. What most counters do is linear counting. What I am working on and what some one I now does is multidimensional counting. This involves gathering multiple counts that can be combined in a relative form before hand to be solved as linear equations exponentially growing your advantage. My view of what Tarzan does is at least a 3 dimensional counting system. The potential for advantage is staggering. A linear count confines the potential for advantage to a straight line. A 2 dimensional count spreads the information out into a plane allowing much greater detail. A 3 dimensional count spread it out to 3 dimensions further increasing the subtle details. This has its limits but I hope you get the idea.

 

[Automatic Monkey]

This will probably go over your head but I will try anyway. What most counters do is linear counting. What I am working on and what some one I now does is multidimensional counting. This involves gathering multiple counts that can be combined in a relative form before hand to be solved as linear equations exponentially growing your advantage. My view of what Tarzan does is at least a 3 dimensional counting system. The potential for advantage is staggering. A linear count confines the potential for advantage to a straight line. A 2 dimensional count spreads the information out into a plane allowing much greater detail. A 3 dimensional count spread it out to 3 dimensions further increasing the subtle details. This has its limits but I hope you get the idea.

It doesn't get you very much. Tarzan does as he does, it works, and far be it from me to tell him what to do, but I'm telling you it will not put significantly extra money in your pocket relative to using any of the standard level 2 counts.

What we can do with counting is limited by the normal distribution of the removed cards. Sure, we can devise a multiparameter count that will predict what will happen when a bunch of 4's and 6's are removed and the 5's and 3's are left behind, but its utility is limited by the infrequency which with this will happen. Other than a bizarre Wonging scheme where we abandon shoes when the 3's and 5's get dealt out, we have no way around this. The standard counts allow for a generous range of non-uniform removal distribution while still allowing us to predict an advantage and select a correct play.

If you're looking for advice on how to get the money with counting, here is my advice:

1) Learn one of the following counts:

• HO2 with ace sidecount (included because a good ace-neutral count is useful in some situations)
• RPC (used by a lot of serious shoe grinders)
• Mentor (slightly better parameters than RPC)
• Zen

2) Develop and practice ways to:

• Hunt for good penetration
• Exit and enter shoes, quickly and efficiently as your advantage changes
• Backcount multiple shoes, alone and with partners
• Speed up your play, by selecting fast games and making/keeping them fast
• Beat sidebets
• Maximize comps
• Exploit dealer errors
• Use scavenger play to exploit your civilian tablemates.

Look, I'm just a monkey, maybe what you're saying is over my head too. These are just suggestions that have worked for me.

 

[blackjack avenger]

a matter of pedigree?

You guys want to keep going back to your computer generated simulations as the definitive answer to this equation. These results do not always translate into the real world of blackjack play as I have said on numerous occasions. You need to think outside the box a little.

Lets go back to an example I used earlier, that was first brought up by 21forme. Counting 2 sometimes 3 tables at once. I almost always am tracking at least one other table besides the game I am currently playing. If the count on the other table heads north I am on that table in a nanosecond. This basically gives me double the high count percentages, which your simulations don't show. This probably boost my overall win rate 50 to as much as 100 percent. So while your complex counts may boost your earnings 5-10% as simulations show, you probably aren't able to track several tables at once using a level 3 count like halves or some count with a side count or several side counts. And this is just one example of some things you are free to do by keeping things simple that you may not be able to do with a more complex count. In the real world, everything doesn't fit so nicely into a computer simulation.

As I have said on previous occasions, the discussion isn't about the math. The discussion is about whether it is worth the effort and if there are better ways to go. If you take the discussion outside the computer and into the real world, you will find a different answer. And as I said in another thread, which has since been removed, I find it very interesting that many long time, very experienced players, several who play exclusively for a living have come to the same conclusion. (I am not going to list names again.)

What makes u think a halves user can't do everything you said?
ST, multi table etc. Let me assure you one can.
In fact the more hands a hi lo and halves players get in the more halves outpaces hi lo. One of the strengths of halves is more positive hands played! If someone resizes their bank on wins and losses, the advantage of halves DOUBLES over hi lo.

You base your thoughts on pedigree of others? Nothing against anyone you mention.
I will go with:
Snyder writes of how small increases add up with bank resizing.

Schlesinger halves "Cadillac of counts" do to its str. across games.
Also, in BJA3 Sch shows why in a team setting a superior count player should earn more!

Qfit in modern bj halves is the top count without side count

Wong "halves makes almost no error in estimating advantage"
"You can make a million with hi lo, with halves you will make it faster"

It's amazing in this mad world a hi lo sim is gospel but a halves sim is questioned by those who have probably never used the count! The burden of proof should be on lower level counts to prove themselves because the sims don't back their users "subjective" claims.

If someone can't test out on a higher level count or use it easily then they shouldn't use a higher level count.

Shouldn't you & I be railing against threee :laugh: and his side counts?