gazman: Level 2 count

[stainless steel rat]

> To be honest, I didn't suspect you of
> anything, and I apologize for coming out of
> nowhere and blasting you randomly. You were
> saying that a lot of people in the blackjack
> community suspect others of duplicity; I
> don't know anything about that. It's really
> just a personal problem I have: I love
> dismantling non sequiturs and arguments
> based on specious reasoning.

> I understand now that what you were saying
> made perfect sense to you, but the
> progression of your responses wasn't making
> sense to me, and I'm sure it wasn't making
> sense to most anyone else either. Which is
> probably why you got blasted by Don. Let me
> show you how it looked from the perspective
> of someone who is not inside your head:

> 1) gazman: "Question: is an L2 much
> better than an L1?"
> 2) SSR: "No; you only get 3 extra
> correct plays per 100"
> 3) Myooligan: "Don't compare that;
> compare SCOREs to get the real answer."
> 4) SSR: "OK, I compared winrates in
> CVCX, and Hi-Lo eeked out a better one than
> Zen."

> Now, for anybody reading your responses,
> this is where things went awry. You posted
> stats that demonstrated that the L2 Zen
> performed poorer than the L1 Hi-Lo, so at
> this point, it seems like you're response to
> the question "is an L2 better than an
> L1" is that for some DD games, an L2
> will underperform an L1. NOWHERE in that
> post did you mention that that was true only
> if you did the comparison with a fixed,
> non-optimal bet spread, and nobody even
> ASKED if an L1 outperforms an L2 when you do
> such an improper comparison.

> So it is assinine to think that anyone could
> magically figure out the comparison you were
> actually trying to make. If your response to
> Myooligan had been "OK, Zen SCOREs are
> clearly better than Hi-Lo SCOREs; BUT,
> SCOREs assume you are playing with an
> optimal bet spread, and since you generally
> can't do that, situations can arise in which
> Hi-Lo will outperform Zen," then what
> you were saying would have been totally
> clear. But instead, you cited situations in
> which Hi-Lo outperforms Zen with no preface
> at all, and you were making absolutely no
> sense to anyone.

> Now:

> 5) Don: "No, that's incorrect; SCOREs
> show that Zen is superior."
> 6) SSR: "When I post numbers I _always_
> post numbers that reflect what I am
> playing."

> So now do you understand why everyone, Don
> included, was scratchin their head about
> your post? Previous to post 6), you never
> mentioned that you were doing an improper
> comparison.

Note that I didn't consider this an "improper comparison" myself. In other contexts I would not begin to argue the point however. I probably go too far to the empirical side of things, that is to say, if for a circumstance I am interested in, something works, then I will say that it works, and do my best to describe the circumstances. whether it works in all other circumstances I rarely bother about. There are so many variations to this silly BJ game, that I won't try to learn the best betting ramp and spread and min bet size for every possible game, because there are a huge number of variables that can change, penetration being a good example.

I've been playing BJ a long time. I've been a computing professional for about 7 times longer. And I certainly realize that there is a lot that I have yet to learn about communication with other BJ players. Probably my biggest shortcoming here is that I have now been playing pretty successfully for a little over five years now, starting the clock at my first casino visit after several months of preparation and practice with hi-lo. Until maybe 3-4 months ago I had no idea there were actually forums where card counters / APs met to discuss things, as it all seemed like something that would be pretty ridiculous to discuss where it can be seen by all, and traced to your desktop very easily. But here we are, and while I certainly understand the game, and my counting system, about as well as anyone understands their system, it is pretty apparent that I have a ways to go in picking up on the various technical terms that are used by many, but probably not understood very well by most. I sort of "joined the party pretty late" and as a result, words and terms usually end up meaning something different than they did to my circle of friends that learned to count among other things...

All I can say is that I'll do my best to get up to speed on the terminology, and on "current practice and meaning" of various topics, so that my word XXX means the same as what it means to you and others...

Remember that this is a "beginner's" page. I certainly consider myself a "beginner". Probably not in actual card counting skill, but certainly in talking to other BJ players that do not know me personally. Hence a wide "communication gap" at times.

> And see, man, that's how a lot of your posts
> tend to be. They probably make a lot of
> sense to you, but to most others, it's just
> a lot of circumlocution and subject
> switching. And the way that you went awry in
> this thread is the way that you've gone awry
> in many other threads as well. If you didn't
> continuously come so STRONG in your posts,
> it probably wouldn't be such an issue, but
> you do, so.

I certainly didn't realize that I "came on strong". It was never an intention. I really came here to learn whatever might be exposed (if anything) and at the same time, offer advice to those that appeared to be able to use the advice I have to give...

> Hey, maybe you don't realize that you're
> doing it, but that's where feedback comes
> in, right?

yep... don't mind feedback. Anyone in academia that publishes papers, comes up for promotion and tenure decisions, and applies for federal grand funds has to be used to feedback, positive and negative. Probably the important point is what one _does_ with that feedback.

My goal is to become as good at counting as I am at computing. However, there are some things I don't believe I have to master in order to be successful at counting, based on past experience. CVCX has given me a tool to use to figure out how to beat a particular game, without having to resort to books, to SCORE, or anything else, which fits me just fine here. I use "books" every day. My office is full of them (probably about 1500 books on shelves here). BJ is a recreational thing I do to get away with my wife, but it is also something profitable at the same time. I hope to keep it in that limited perspective. And to get along with folks here if at all possible, or if not, to continue playing as I have for the past five years anyway, since it has been working.

[pm]

> I've been playing BJ a long time. I've been
> a computing professional for about 7 times
> longer. And I certainly realize that there
> is a lot that I have yet to learn about
> communication with other BJ players.
> Probably my biggest shortcoming here is that
> I have now been playing pretty successfully
> for a little over five years now, starting
> the clock at my first casino visit after
> several months of preparation and practice
> with hi-lo. Until maybe 3-4 months ago I had
> no idea there were actually forums where
> card counters / APs met to discuss things,
> as it all seemed like something that would
> be pretty ridiculous to discuss where it can
> be seen by all, and traced to your desktop
> very easily. But here we are, and while I
> certainly understand the game, and my
> counting system, about as well as anyone
> understands their system, it is pretty
> apparent that I have a ways to go in picking
> up on the various technical terms that are
> used by many, but probably not understood
> very well by most. I sort of "joined
> the party pretty late" and as a result,
> words and terms usually end up meaning
> something different than they did to my
> circle of friends that learned to count
> among other things...

> All I can say is that I'll do my best to get
> up to speed on the terminology, and on
> "current practice and meaning" of
> various topics, so that my word XXX means
> the same as what it means to you and
> others...

> Remember that this is a
> "beginner's" page. I certainly
> consider myself a "beginner".
> Probably not in actual card counting skill,
> but certainly in talking to other BJ players
> that do not know me personally. Hence a wide
> "communication gap" at times.

Hey, I definitely consider myself a beginner as well. The only reason I said anything at all is because the communication gap (or whatever) was apparent to me, and I've seen the gap occur before on several occasions, so I figured, what the hell, I'll point it out this time.

But I see where you're coming from; I tend to stay focused only on my thoughts as well (probably why I also end up getting blasted by Don every now and again).

[Wolverine]

I've tried it in my head but find during practice sessions that I will miss the actual RC by 1 or 2 integers. Not fatal, but certainly not the accuracy I have grown accustomed to just counting the RC without the aces. When I count the aces outside of my head, I keep the RC very well and rarely blow the ace side count (but I will let it go rather than lose the RC). Typically, I use my chips to help keep the count of aces. In this forum, I'm not sure I want to share *exactly* how I do that.

How would you recommend tracking the aces besides mentally (which is my goal)?

Now that I know tracking the aces isn't needed for a betting tactic, I'm really not sure I plan on tracking aces until I can count them mentally and not screw up the HiLo RC.

[stainless steel rat]

> Hey, I definitely consider myself a beginner
> as well. The only reason I said anything at
> all is because the communication gap (or
> whatever) was apparent to me, and I've seen
> the gap occur before on several occasions,
> so I figured, what the hell, I'll point it
> out this time.

> But I see where you're coming from; I tend
> to stay focused only on my thoughts as well
> (probably why I also end up getting blasted
> by Don every now and again).

OK. Then caio until the next long thread comes along, probably at my urging.

[Myooligan]

> If it is the "defacto standard" that we always > compare two counts on a hyper-theoretical plane

But check it out, Stainless. . . I think the people who have been challenging you throughout this thread would disagree that the power of Zen, compared to Hi-Lo, is merely theoretical. We're arguing, in many different ways that I'm not going to go back and quote, that you're using faulty logic. . . and if the rest of the crowd is like me, the main reason they even bother posting is because they want you to get the best SCORE possible! And whether you overlooked something in the settings in that CVCX sim, or you truly found an instance where Hi-Lo outperforms Zen, I think it is still to your advantage to get a stronger grasp on the SCORE concept.

But to address your sim in particular: I don't have CVCX, but I've tinkered around with the trial version. Try going back to the same sim, set identical bankrolls, and then select "Optimize For Bankroll Growth" under the Options tab. I believe you'll get different results. Or, it could be that the risk/bankroll levels you've selected are idiosyncratic: If the advantage happens to be exactly .5, 1, 1.5, 2.0, 2.5 for Hi-Lo TCs 1-5, for instance, then you end up getting close to true kelly betting. But I'll bet if you scroll through various bankrolls, repeating the process described above, you'll find that Zen generally comes out ahead, by more than a nickel or two.

Please let me know what you find, if you end up trying that.

-Myoo
ps How'd you do at that DD game? =)

> I hope I explained what I was doing. I tried
> this multiple ways.

> 1. Set a specific minimum bet, nothing else,
> and then used the two distributed CVCX sims
> (one for zen, 2B rounds, one for hilo, 2B
> rounds) and gave the expected win rate per
> hour.

> 2. Set a specific BR, and did the same.

> 3. set the BR to a point that each count had
> an exact 10% ROR and gave the BR requirement
> for each.

> The point was _never_ that HiLo was better
> than Zen. Didn't expect it to be, didn't
> claim it was. Don't even know how or why I
> would want to claim that never having used
> or studied the Zen count.

> The point was that for the specific game I
> play, it appears that hilo works a little
> bit better, regardless of the score value
> that says zen is better, period. Did I claim
> HiLo works better for all games? Nope. For
> all penetrations in my game? Nope. For all
> bet ramps? Nope. For all bankroll sizes?
> Nope. For all ROR requirements? Nope.

> If you have CVCX, you should be able to
> produce _exactly_ what I produced without
> running a sim at all. Someone else here
> reported that they found my numbers to be
> accurate for the conditions I claim.

> I personally don't care about
> "theoretical comparisons" that say
> "for all games, penetrations, etc, this
> count is better." Because that leaves
> the "out" that perhaps it is
> better in all the games I do _not_ play, and
> worse or equal in the games I _do_ play. I
> really care about what I play, not what I
> _could_ play.

> If it is the "defacto standard"
> that we always compare two counts on a
> hyper-theoretical plane, that's fine. But
> what counts for me is what happens when I
> walk up and sit down to a real game, not
> what might happen when I walk up and set
> down to any possible game...

> I believe that is where this thread took a
> wrong turn, because I never intended to say
> nor imply that hilo is better. I said
> "it is good enough and easier, and the
> potential gain from a more complex and
> hard-to-use count doesn't appear to be that
> significant to me." Others pointed out
> that complex systems lead to more errors for
> casual users, making the comparison of a
> pure theoretical outcome even less
> realistic.

> I know what I ought to be able to do
> "in theory" for various things.
> And I also know what I can actually do
> "in practice". Only rarely do the
> two reach equal ground, whether it be how
> far you can drive a golf ball, as opposed to
> how far you can drive it and keep it in
> play, to how fast your car will to as
> opposed to how fast it can go without your
> ending up in jail or in the morgue.

> There's plenty of room for practicality,
> IMHO.

> Now pardon me while I go play some DD at 66%
> pen with hi-lo and win some money. Without
> getting a headache in the process.

[stainless steel rat]

> But check it out, Stainless. . . I think the
> people who have been challenging you
> throughout this thread would disagree that
> the power of Zen, compared to Hi-Lo, is
> merely theoretical. We're arguing, in many
> different ways that I'm not going to go back
> and quote, that you're using faulty
> logic. . . and if the rest of the crowd is
> like me, the main reason they even bother
> posting is because they want you to get
> the best SCORE possible! And whether you
> overlooked something in the settings in that
> CVCX sim, or you truly found an instance
> where Hi-Lo outperforms Zen, I think it is
> still to your advantage to get a stronger
> grasp on the SCORE concept.

> But to address your sim in particular: I
> don't have CVCX, but I've tinkered around
> with the trial version. Try going back to
> the same sim, set identical bankrolls, and
> then select "Optimize For Bankroll
> Growth" under the Options tab. I
> believe you'll get different results.

I did this. In fact, someone suggested that the Zen has a lower ROR for a given bankroll. To test that, I set both sims (hilo full indices, zen full indices) let cvcx choose the best betting ramp and min bet, and then the only value I changed was the bankroll. I adjusted the starting bankroll for each until I got a ROR of exactly 10%. Hi-Lo had a _lower_ BR for this specific game (DD, 66%, etc). I have no idea why, but for this specific game it seems that Hi-Lo is better. Is it serendipity that I picked the right game and the right count? Must be as I certainly didn't start playing by scouting the games, then find the game I liked best and then choosing the best counting system for that game.

I certainly realize that in general, a L2 count is better at PE than a L1 count, and I understand why as it models the card removal effect more accurately. Is it unfortunate that I happened to pick my favorite game, which just happens to favor hi-lo over zen somehow? Of course. Had I specifically set out to find a particular set of rules where Hi-Lo was better, I'd consider that a bit dishonest in this discussion. But I didn't. And, in fact, it seems that all the discussion about "you can't compare with sub-optimal betting ramps, and sub-optimal minimum bets" is simply wrong here. Anyone with CVCX can look at the zen 80 and hilo sims and make this comparison and I can't find any case using my exact DD rules where Zen is better. That clearly doesn't mean it isn't. And that wasn't my point.

The only loose point here that I am going to try to solve later today is that I don't know the exact details about how the sims were run by Norm, so I'm going to run them again, making certain that each is run with everything but the counting system set identically. If Hi-Lo wins again, good. If not, also good. I didn't pick Hi-Lo because it was the best. Had I done that I would have used what a particular math genius college professor of mine did and used the exact card-removal-effect numbers to maintain a real count of exact advantage. But I'm not sure I could cope with adding/subtracting 2-3 digit fractional decimal numbers, whereas he had a gift for that that was beyond belief. I picked Hi-Lo because I thought I could learn to do this count accurately, quickly, and with little effort during a playing session. Whether I could learn a L2 (or higher) counting system, I don't know. I suspect I could, but how easy it would be to use I don't know. Here I listened to many others that said "L2 is fatigue-prone".

As I said, I had no agenda, and was _really_ surprised when for this game Hi-Lo popped out a higher (not by much) win rate, a lower ROR for the same BR, etc. I'll post some details on the sims after I get them running later today.

>Or, it
> could be that the risk/bankroll levels
> you've selected are idiosyncratic: If the
> advantage happens to be exactly .5, 1,
> 1.5, 2.0, 2.5 for Hi-Lo TCs 1-5, for
> instance, then you end up getting close to
> true kelly betting. But I'll bet if you
> scroll through various bankrolls, repeating
> the process described above, you'll find
> that Zen generally comes out ahead, by more
> than a nickel or two.

See above. Someone asked that question yesterday and when I looked the numbers stayed in hi-lo favor for this specific game. I left the betting ramp at "optimal" for each, let the min bet float (both went to 10 by the time penetration hit 60%) etc...

> Please let me know what you find, if you end
> up trying that.

> -Myoo
> ps How'd you do at that DD game? =)

I have had good luck. My best ever playing session was back in the summer at this very game, dealer was hand-setting the cut card at 75%, there were two of us playing, myself starting at $25-$150 or so, another player spreading black $100-$1200. After 30 minutes at a continual <=0 count, I had probably lost one hand, and was up a grand or so. Decided to start playing black and spread 1-8 there. No heat whatsoever, and when I checked out, I was up just over $9500 dollars while the big black player was down a small fortune (he was not counting by any system that would approximate the betting ramp for hi-lo, he seemed to be doing some sort of oddball progression approach. He lost way more than I won, so we never needed any chip fills and the chips never seemed to drop off much in the dealer's tray.

That's my personal record. I have also dumped $1200+ in 30 minutes playing $5-2x$25 or so, with monster counts that turn into a nightmare on elm street. That's one reason I like DD, monster counts don't kill me nearly as often as they do on 6d shoes. I remember "the shoe" in vegas at the grand, where the RC hit +60 by the middle of the third deck, I was playing $5-$100, and with my wife watching, I played 18 consecutive rounds with a huge TC, betting $100 each time, and losing every last hand. I know it was 18 consecutive losses because my wife was standing there watching and she didn't let me forget it quickly.

[Don Schlesinger]

> I did this. In fact, someone suggested that
> the Zen has a lower ROR for a given
> bankroll.

It does, without question.

> To test that, I set both sims
> (hilo full indices, zen full indices) let
> cvcx choose the best betting ramp and min
> bet,

Those min bets were different, right?

> and then the only value I changed was
> the bankroll. I adjusted the starting
> bankroll for each until I got a ROR of
> exactly 10%. Hi-Lo had a _lower_ BR for this
> specific game (DD, 66%, etc).

Must have had a lower win rate, too, no? So, what are you comparing??

> I have no idea
> why, but for this specific game it seems
> that Hi-Lo is better.

Not a chance. You're doing something wrong.

Don

[stainless steel rat]

OK. I ran a sim for Hilo complete and Zen complete. Same game, DD, DAS, S17, LS, 66% pen, I let CVCX compute everything else optimally, with the "optimize bankroll growth" option set.

Results:

1. Hi-Lo. $37.19 per hour win rate. c-SCORE 111.07, ROR 16.6% with a $3000 BR, spreading 1-8. CVCX chose an optimal bet size of $10, betting $10 at TC<=0, $15 at TC=1, $30 at TC=2, $45 at TC=3, $65 at TC=4, $80 at TC>=5.

2. Zen. $34.77 per hour win rate. c-SCORE 98.57, ROR 18.2% with a $3000 BR, spreading 1-8. CVCX chose an optimal bet size of $10, betting $10 at TC<=1, $15 at TC=2, $25 at TC=3, $35 at TC=4, $40 at TC=5, $50 at TC=6, $60 at TC=7, $75 at TC=8 and $80 at TC>8.

For this specific game, for some unknown reason, HiLo simply out-performs Zen. Why, no idea. Did I do anything wrong in _this_ pair of sims? I doubt it. I set the rules _identically_. I did not run sims for non-DD games, where for all I know Zen will beat HiLo badly.

If someone wants to repeat this, here is what I did in CVCX:

1. start CVCX.

2. click "new sim"

3. attach a strategy, either Snyder's Complete Zen, or Wong's Complete Hi-Lo.

4. true count calculation = round.

5. TC resolution for 1,2 decks = 1/4 deck.

6. Dealer actions Dealer peeks on 10/A.

7. Hard DD on any hard hand.

8. Soft DD on any soft hand.

9. Max cards to DD on 2.

10. Max split to 4 hands.

11. no resplit aces, no multiple draws after Ace split.

12. click "sim page"

13. Included rules S17 DAS LS.

14. Rounds 100M

15. DD only, one player (heads-up)

16. click "go".

Only difference in two sims is the playing strategy.

Don't know what more I can say here. If someone comes up with different numbers than mine, we need to figure out who is doing something wrong. If you want to compare different games, I can even run that.

so...

[stainless steel rat]

> It does, without question.

> Those min bets were different, right?

No. Both jumped to $10 at somewhere around penetration of 58-60%....

> Must have had a lower win rate, too, no? So,
> what are you comparing??

See my just finished sim data in this sub-thread. Hi-lo for _this specific game_ has a higher win rate. I ran the sims myself. I gave the _exact_ settings within CVCX used to produce the two sims. While I could have done something wrong, I certainly don't see what, and since I gave every mouse click in the post I just finished, someone ought to be able to say either "ok, here is your mistake" or "ok, for this game hilo is simply better".

> Not a chance. You're doing something wrong.

Please look at the post adjacent to yours in this sub-thread and tell me what I am doing wrong. I'm all ears.

> Don

[stainless steel rat]

[stainless steel rat]

I just completed two sim runs, one for hilo-full and one for zen-full, this time one billion rounds each, everything the same as before except that I went to two players at the table rather than heads-up to see if that would change anything.

HiLo: Hourly win rate: $35.45, bet $10-80, etc.

Zen: Hourly win rate: $34.54, bet $10-$80, etc.

Only constraints I supplied were bankroll ($3000) and bet spread (1-8).

make of it what you will, but this time there is nothing that I am doing "wrong" here. This particular DD setup just seems to be hi-lo's "game".

[Don Schlesinger]

Very quickly, using BJRM, I ran the two side-by-side (I18 only).

Zen won a little more and had a 2% lower ROR.

I repeat: Something is wrong.

Don

[stainless steel rat]

> Very quickly, using BJRM, I ran the two
> side-by-side (I18 only).

> Zen won a little more and had a 2% lower
> ROR.

> I repeat: Something is wrong.

> Don

The question is "what"? Clearly not my setup here, now, as I very carefully made everything the same. Perhaps it is the parameters for the "snyder full-zen" as supplied with CVCX? I checked the hi-lo indices and found nothing untoward. Don't know enough about Zen to see if the CVCX parameters omit anything important or what.

I'll try an I18 run to see if that produces your results, since I have been running with full indices...