BJInfo Open Source UBZ II V0.5

[boneuphtoner]

Here is a comparision between Mentor and Zen (canned sims) by Qfit.

Yeah, if you use the canned sims of Zen, it will look inferior to just about any level 2 strategy. I think Norm told me he ran those with the true edge methodology. Which results in lowered SCORES. When I was debating about which system to choose, I almost went with count per deck Zen. In my sims, using TC per deck, it does indeed outperform RPC and Mentor by noticeable margins. That mirrors what Norm and Cacarulo have reported. Cacarulo also reported that Zen outperforms Halves. In my sims, Zen barely edged out Halves, only to yield substantially to Halves at deep penetrations.

 

[boneuphtoner]

The optimized composite UBZ is the easiest of the level-2 systems without any compromise of performance. Now who is or will be using it or a close facsimile thereof.

Me! I've decided to switch (for now) but haven't employed it in a casino YET! I came up with an extremely powerful version of Zen in early summer. I used SBA generated risk averse indices and I employed the ZG 60 indices. And I used it successfully out in Vegas this summer. I could fly through the level 2 values at under 15 seconds with perfect accuracy every time. I even got it 12 seconds once. But I certainly was never confident of my deck estimation skills. What bugged me even more was the fact that my TC conversions became even more important when the count got really high when you size your bets and yo have to do yet another TC conversion before you took insurance or changed up the play of your hand. The level 2 values didn't phase me at all, it was the TC conversions that tired me out. And this was all while attempting mostly 6 deck shoes. The thought of having to use different conversion factors for 8 deck shoes also seemed daunting to me....all the while not feeling at all confident that my discard tray estimates are even remotely close.

Initially, I never thought about UBZ as my canned CVCX sims showed that it was an inferior count even compared with some level one counts when employed with shoes. But then I decided to try my own RA indices and use all of the catch 22 plus a few extra that I knew were easy to remember and fell inline with certain decisions (for example, you use the same indices for splitting 10s v. 5,6 as you would splitting 9s versus 7 and Ace, as well as doubling down on your soft 20s versus 5 and 6). I wanted a single set of indices that could be used for all decks, so I ran my RA indices using 4 decks as the compromise. It worked beautifully....besting all of the level one counts (balanced and unbalanced) and level 2 counts with the sweet 16 indices. And holding its own against the level 2 counts (Mentor and RPC) with FULL UNcompromised indices. It was a slightly weaker with single deck games, but I found that replacing all of the plus 10 indices with +5 did the trick...again, that change is for single deck only.

Right now, the true counting I'm doing with this is to determine when to wong-out (more discussion on that in the next email). I do fudge a little as I approach the pivot point of zero, being slightly more aggresive in the early part of shoes, less agressive later on, but that is about it. Even those subtle fudging changes would increase my performance beyond what I've already demonstrated. I like the idea of not worrying about performing another TC conversion when deciding whether to take insurance or change up the play of my hand when I have a big bet out and the adrenaline is flowing. I may change my mind and go back to a balanced count one day, but for someone who only has the opportunity to play 50 hours per year (even that is generous since I don't live near any casinos) this approach just makes sense. My sims show great performance without true counting so I don't even need to go there is I don't want.

 

[Mimosine]

if the "canned sims" you are referring too used the original UBZII indices it is no wonder that our intuitive index numbers bested it, in our first sims.

as far as wongin and wongout points, the ones i provided on page 2 or 3 are based on an RC to TC conversion as a function of the number of decks dealt. They should as accurate as any wonger would need....

 

[zengrifter]

as far as wongin and wongout points, the ones i provided on page 2 or 3 are based on an RC to TC conversion as a function of the number of decks dealt. They should as accurate as any wonger would need....

Revere, in his private system booklets and lessons, taught betting by RC, not TC. He had RC betting schedules that were calibrated to each 1/2D, like this:

(decks remaining - count - bet)
3-3-2
3-5-4
3-7-4+4
---------
2.5-2-2
2.5-4-4
2.5-6-4+4
etc.
​

I used to have my betting schedules inscribed on a lighter (late 70s), and before that on a slip of paper I would slip into the side of a cigarette pack.

"Betting by RC is more accurate." -Revere.

This type of betting is better suited for today's RC systems than for the TC systems of the 70s. zg

 

[boneuphtoner]

as far as wongin and wongout points, the ones i provided on page 2 or 3 are based on an RC to TC conversion as a function of the number of decks dealt. They should as accurate as any wonger would need....

Yeah, I saw those...I'd like to discuss further.

For aggressive wonging, I think your numbers look great. But I use a semi-wonging approach....rarely do I have a situation where I can wong out and find a new shoe in a reasonable time. That is, I play until it gets sufficiently negative that the count is unlikely to recover. According to Blackjack Attack, for balanced level one strategies, that is around -2 for most of the shoe. According to CVCX, that true count is equal to about a disadvantage of approximately 1.0%. For level 2 counts, this same disadvantage is equal to a TC of -3. Consequently, for a six deck shoe, this is what I get for wong-outs using this strategy:

1 Deck Played -35
2 Decks Played -28 round to -25
3 Decks Played -21 round to -20
4 Decks Played -14 round to -15

What do you guys think?

 

[Mimosine]

For level 2 counts, this same disadvantage is equal to a TC of -3. Consequently, for a six deck shoe, this is what I get for wong-outs using this strategy:

1 Deck Played -35
2 Decks Played -28 round to -25
3 Decks Played -21 round to -20
4 Decks Played -14 round to -15

What do you guys think?

well these were my originals... mind you I only spent about 45 minutes cooking them up:

WONG OUT
5 Decks remain (1 deck played) RC = -28 • TC = ~ -0.8
4 Decks remain (2 decks played) RC = -24 • TC = ~ -1.0
3 Decks remain (3 decks played) RC = -18 • TC = ~ -1.0

so they aren't that far off from yours, i just prefer something closer to -1.5 rather than -3, TC To wong out. if after 3 decks, and you are still at a negative count, I rarely stick around for a 4th, unless the pen is at 85-90% (and i have a few dealers that give me that). I do think your numbers look good, i only question the logic of waiting around for 66% of the cards to be dealt before you wong out.... i realize sometimes it is hard to table hop, but if you're playing to 66% pen on a 75% dealt game, then really you are in a play-all environment and the value of wonging out decreases to almost nil i would imagine. maybe.... :/

what do you have for wongin points? do you wong in at TC =+1.5 and bet 2 units? that's what I was thinking for the wong in points....

 

[nightspirit]

Haven't had the time to check the wongout points, but here are a few bet ramps for 2 and 6 decks. When penetration changes, they change too. So this is just the first round. H17 must also be integrated. These are optimal bet ramps, but I'm sure we can make a few simplifications here, too.

I choosed the same spreads as presented in BJA. The vertical column to the left shows the units we should bet at the RC's, shown in the center of the table.
(Dead link: http://imageshack.us) _

_

 

[boneuphtoner]

i realize sometimes it is hard to table hop, but if you're playing to 66% pen on a 75% dealt game, then really you are in a play-all environment and the value of wonging out decreases to almost nil i would imagine. maybe.... :/

Actually, in evaluating a couple of different level 2 balanced counts where I consistently wonged out at TC -3 (equivalent to a TC of -2 in HiLo), the SCOREs were almost exactly halfway between aggressive wongout at a TC of 1 and a play-all approach. I don't generally table hop...I generally sit down at the shuffle and take bathroom, cell phone breaks once I reach my wong-out point. And again, my CVCX sims show that this approach puts me about halfway between aggressive wonging and play all approaches.

what do you have for wongin points? do you wong in at TC =+1.5 and bet 2 units? that's what I was thinking for the wong in points....

I haven't tried this yet in a casino, but I would probably wong in close to the pivot of zero (RC).

 

[nightspirit]

The QFIT sim above is in contradiction to Caraculo's ranking, Mentor outperforming ZEN. zg

Boneuptoner mentioned that true edging is the reason for the bad performance of Zen in the comparision above. So I ran a sim using the True-Count-Indices presented in the library of BJFO. True count was calculated with half deck resolution for 6 to 2 decks, and quarter deck resulotion for the last 2 decks. Now, the ranking like Cacarulo mentioned is rearranged.
(Dead link: http://imageshack.us) _

_
(Dead link: http://g.imageshack.us/g.php?h=530&i=zentcvmentordt1.jpg) _

_

 

[zengrifter]

Boneuptoner mentioned that true edging is the reason for the bad performance of Zen in the comparision above. So I ran a sim using the True-Count-Indices presented in the library of BJFO. True count was calculated with half deck resolution for 6 to 2 decks, and quarter deck resulotion for the last 2 decks. Now, the ranking like Cacarulo mentioned is rearranged.
(Dead link: http://img114.imageshack.us/my.php?image=zentcvmentorxu5.jpg) _

_

Which is the red and which the green? zg

 

[nightspirit]

Which is the red and which the green? zg

fixed

I already have the deck adjusted wongout and wongin points for 6D done.

Here are my initial estimates, again we could try to round them.

6D IRC = -24

WONG IN

5 Decks remain RC = -8 • TC = ~1.20 (bet 2units)
4 Decks remain RC = -6 • TC = ~1.25 (bet 2units)
3 Decks remain RC = -4 • TC = ~1.33 (bet 2units)

WONG OUT

5 Decks remain RC = -28 • TC = ~ -0.8
4 Decks remain RC = -24 • TC = ~ -1.0
3 Decks remain RC = -18 • TC = ~ -1.0

I just tried to find out how you have calculated this Wongin- and -out points, since i got a others, maybe you could shed light on this topic.

I reread an old post of Brett Harris on the bjmath site and one of Knox, which he made here some weeks ago and where he explained how he tc'ed KO. So I applied this technique for UBZ but I'm not sure if a true count of -1 / +1 for a level 2 system is equivalent to the same of a level 1 system like hi-lo, i think you integrated that somehow.

This is my calculation:
First, thats how the RC would be, if the decks would stay neutral throughout the shoe, since the unbalanced per deck is +4 for UBZ:

(IRC=-24)
decks rem./neutral RC

6 / -24
5 / -20
4 / -16
3 / -12
2 / -8
1 / -4

Now, when we have have 5 decks remaining and the running count is -25, we would have seen 5 extra high cards. Our true count would be -5 / 5 =-1
(Unbalanced TC =(difference from RC to neutral deck level)/decks.remain.)) and we would wong out. The same applies for a RC of -20, when there only 4 decks left. With other words: we would wong out when we have the number of decks remaining less than our neutral deck level.

hence my wongout points would look like this:
decks rem. / RC / TC

5 / -25 / -1
4 / -20 / -1
3 / -15 / -1

For the wongin-points I used the same concept with reverse logic and these would be my wongin-points for a TC of +1:

decks rem. / RC / TC

5 / -15 / +1
4 / -12 / +1
3 / -9 / +1

When I compare them with yours, yours look more aggressive for wongin, but your TC might be higher according my method.

Consequently, for a six deck shoe, this is what I get for wong-outs using this strategy:

1 Deck Played -35
2 Decks Played -28 round to -25
3 Decks Played -21 round to -20
4 Decks Played -14 round to -15

What do you guys think?

We both used the same type of calculation, only that you used a TC of +3. Now I got it! :grin: thx! I will try to set up a sim during the next week to compare all methods, then we could discuss this further.

 

[Mimosine]

fixed
I reread an old post of Brett Harris on the bjmath site and one of Knox, which he made here some weeks ago and where he explained how he tc'ed KO. So I applied this technique for UBZ but I'm not sure if a true count of -1 / +1 for a level 2 system is equivalent to the same of a level 1 system like hi-lo, i think you integrated that somehow.

it isn't the same. i talked with Fred Renzey about this quite a bit. In essence it is a level two 1/2 rank count. Fred told me to look at his Kiss III count which is level one 1/2 ranked. both Kiss III and UBZ set the pivot point = to TC = +2.

UBZ does this differently than Kiss III thought. that is where the confusion started....

fixed
This is my calculation:
First, thats how the RC would be, if the decks would stay neutral throughout the shoe, since the unbalanced per deck is +4 for UBZ:
Now, when we have have 5 decks remaining and the running count is -25, we would have seen 5 extra high cards.

well technically you would have seen 3 high cards, two Ts and an A.

anywho, the calculation is hard to explain. everything has to be divided by 2 at some point along the way, since it is a level 2 count - for the sake of having something to compare to a level 1....

here are my numbers, Decks remaining on top axis, RC on left axis, TC in center. note that when RC = 0, TC = +2 everywhere in the deck:

	5	4	3	2	1
-24	-0.4	-1	-2	-4	-10
-22	-0.2	-0.75	-1.67	-3.5	-9
-20	0	-0.5	-1.33	-3	-8
-18	0.2	-0.25	-1	-2.5	-7
-16	0.4	0	-0.67	-2	-6
-14	0.6	0.25	-0.33	-1.5	-5
-12	0.8	0.5	0	-1	-4
-10	1	0.75	0.33	-0.5	-3
-8	1.2	1	0.67	0	-2
-6	1.4	1.25	1	0.5	-1
-4	1.6	1.5	1.33	1	0
-2	1.8	1.75	1.67	1.5	1
0	2	2	2	2	2
2	2.2	2.25	2.33	2.5	3
4	2.4	2.5	2.67	3	4
6	2.6	2.75	3	3.5	5
8	2.8	3	3.33	4	6
10	3	3.25	3.67	4.5	7
12	3.2	3.5	4	5	8
14	3.4	3.75	4.33	5.5	9
16	3.6	4	4.67	6	10
18	3.8	4.25	5	6.5	11

 

[nightspirit]

anywho, the calculation is hard to explain. everything has to be divided by 2 at some point along the way, since it is a level 2 count - for the sake of having something to compare to a level 1....

Thank you very much, for the clarification. Now i got it! When I tried to set up a chart this morning I realized that something went wrong with my calculations. Because it's a level 2 count we simple divide the difference of our RC and the supposed neutral level by 2 before we do the TC calculation.

UTC =((difference from RC to neutral deck level)/2)/decks remain.

i.e. RC=4 at 4 decks remain. UTC= (20/2)/4=2.5

So do I get also the +2 when we reach the pivot everywhere in the deck.

Just wanted to understand this stuff before I put it into the simulator.

 

[Mimosine]

NS:

I'm not sure that that table is 100% correct. I ran into some trouble comparing KO, KISS III, HiLo, and UBZII. UBZII is a 1/2 ranked level 2 count, all i know for 100% certain is that the pivot of 0 = TC + 2. The other numbers should be pretty accurate based on my calculations, but there is an assumption that still gives me trouble with this whole thing and that has to do with the NUMBER of cards that are assigned different values. Intuitively I think this presents a problem in the division, i.e. there are four -2 value cards, one -1, four +2, and two +2. I trust Fred's analysis, originally he suggested I just multiply all his Kiss III indicies to get UBZII indicies. this worked pretty well if I recall, but I think our analysis has resulted in a greater return. In any event, when I wong in and out, it is highly qualitative. sometimes the count hits a good TC +1.5 but I wait one more round before I wong in, likewise on wonging out early in the shoe, if the count is only moderately negative I don't leave for a round or two. Sometimes things change.

So overall I think my numbers are pretty good for wonging purposes. Incidentally, they are also the same exact numbers I used to devise my Indicies. As per your sims, they proved to be pretty darn good! So that may support their validity, even if my math and intuition can't.

 

[nightspirit]

I think the table works well. I crossed checked it, dividing all tags by 2. So we get a unbalanced of 2 per deck, which is also our pivot. Now our neutral deck level looks like this:

6 / -12
5 / -10
4 / -8
3 / -6
2 / -4
1 / -2

Now with a running count of -7 at 3 decks rem. we get a TC of -1/3. The same as in your chart with a RC of -14. I haven't read Renzeys book so I hope there is no other magic involved. This is how it should work.
Your intuition was very good! When I looked at the first post of this thread by you, that was your plan:

Initally i'm trying to get these numbers to work for Key Count, Pivot Point, and Advantage:

SD -2, 0, +2
DD -3, 0, +3
6D -6, 0, +6

And I think that really well compares with the indices and the bet ramps I got by CVCX. I have to reran the index generation for SD and use number of rounds instead of the cut card. For 2 decks we could use the "3" version which performed better than the "2". All in all it jived very well with your first assumptions.

 

[nightspirit]

Here is the deck-adjusted sim. I used this great multi-depth feature of CVData to create various betting strategies for the shoe.
Assumed was: 4.5/6, S17, DAS, DOA, LS,SPL4, SPLA1 spread 1-12 (ramp see previous page), 100hands/hour, 3 other players, 2 Billion rounds, with the following indices:

[U]irc	     RC: -12	   RC: -6	 Pivot	      RC: +6	   RC: +10	RC:+12[/U]
12 v. 5	     16 v.10      13 v. 3       12 v. 3	     12 v. 2       X,X v. 5     16 v. 9
12 v. 6		          12 v. 4	 8 v. 6	      15 v.10	   X,X v. 6	
13 v. 2		           9 v. 2	 surrender    8 v. 5		
		           11 v. A	  15 v. 9      9 v. 7		
		           A,6 v.2	  15 v. A      10 v.10		
		           A,7 v. 2	  14 v. 10     10 v. A		
		           A,8 v.5	               A,8 v.4		
		           A,8 v.6				
insurance: +4
irc=-24

and the wongin and -out points according Mimosine’s table:

WONG IN

5 Decks remain RC = -8 • TC = ~1.20 (bet 2units)
4 Decks remain RC = -6 • TC = ~1.25 (bet 2units)
3 Decks remain RC = -4 • TC = ~1.33 (bet 2units)

WONG OUT

5 Decks remain RC = -28 • TC = ~ -0.8
4 Decks remain RC = -24 • TC = ~ -1.0
3 Decks remain RC = -18 • TC = ~ -1.0

1. play-all SCORE 33.59
2. deck-adjusted UBZ SCORE 44.53


Special thanks to Norm!

 

[zengrifter]

What have we got here now? Do we have a single set of indices that are good for 1-2-6-8D, yet? zg

 

[Mimosine]

What have we got here now? Do we have a single set of indices that are good for 1-2-6-8D, yet? zg

no i don't think we're there yet.

i think we're going to have to have deck dependent indicies. we have 6d and 2d, and using your thoughts i think 1d is ripe. (+/- 6, 0, IRC for 6D --- +/- 3, 0, IRC for DD --- +/- 2, 0, IRC for SD).

i mean here is the crux of the matter:

we develop different indicies for each game, then we decide how accurate we want them to be (i.e. 3 index numbers, 4, 6? i think 3-4 is perfect for 6D)

or

we develop one set of index numbers for all - in this latter case we should focus on RC = 0 (TC = +2). Only this number will translate across games, well that and the IRC (which you need for sd and dd 12 v 4-6, 13 v 2-3). By my initial sim called "Moron" (i think ?) didn't perform too well. that was a 6D with all index plays expect insurance being made at RC = 0. I think it was still better than KO though... i'll go look....

well looking at boneuphtoner's posts, i'd say yes:
http://www.blackjackinfo.com/bb/showpost.php?p=52994&postcount=20
http://www.blackjackinfo.com/bb/showpost.php?p=53737&postcount=30

 

[Automatic Monkey]

What have we got here now? Do we have a single set of indices that are good for 1-2-6-8D, yet? zg

No, nor will you. You can probably straddle 6D and 8D pretty well with a single set, but at SD and DD we're too dependent on PE and you really need a separate set for SD and another for DD with any running count system.

 

[zengrifter]

No, nor will you. You can probably straddle 6D and 8D pretty well with a single set, but at SD and DD we're too dependent on PE and you really need a separate set for SD and another for DD with any running count system.

1-2D composite works, as does 6-8D. So the theory is that by adjusting the IRC -AND- mentally doubling the 1-2D composite #s for 6-8D play, we can achieve a single composite index strategy good for any# decks. zg