Quantify how many times per hour what you've described actually happens. No, make that, quantify how many times, every TEN hours, what you've described actually happens in the real world.Originally Posted by bjanalyst
Don
I think the real question is who many times are you making these camouflage plays when the casino is actually looking. The casino is not looking when you are making small bets. They are looking when you are making big bets. And they especially look at insurance plays. If you have a large bet out and HL >= 3*dr then the casino will assume that you should take insurance. In my example, if HL = 6*dr and AA78mTc < -2*dr then Tc < 4*dr and you would not be taking insurance. SD(AA78mTc) = SD(HL) so true count AA78mTc will be less than 2 as often as HL true count is less than two. Also CC(AA78mTc, HL) = 20% so AA78mTc is basically independent of HL count so AA78mTc true count can easily be less than -2 when HL true count = 6. Also insurance is the most frequency strategy change. You put together a great chart in BJA3 that showed the frequency of occurrence of various player hands. I thought that was a great chart. Thanks for the excellent work. I think your BJA3 is an EXCELLENT book and your EoR charts are GREAT.
But there are hard 12 v 2, 3, 4, 5, and 6 strategy changes using AA78mTc with HL which are also contrary to what the casino would expect if the casing were using HL with our I18 to see if the player was a counter or not. I gave examples earlier in this thread where I showed it was better for the player to hit hard 12 v 6 when HL = 3*dr for example if AA78mTc were negative enough. The rule for hard 12 v 6 is to hit hard 12 v 6 if HL + AA78mTc < -1*dr or the slightly more precise rule of hit hard 12 v 6 if HL + 1.5*(AA78mTc) < -2*dr.
Again, the real question is not how many time per hour these camouflage changes come up but what percentage of your large bets, which is what the casino is watching that these camouflage plays come up. When 5m6c is added there is even more camouflage plays for hard 16 v 7, 8, 9 where player would be standing when casino thinks they should hit based on HL and hard 16 v T where player hits with a high HL true count (if 5m6c is sufficiently negative) when casino thinks player should be standing. Casinos know that player should stand on hard 16 v T whenever HL >= 0 and if they see player with a big bet out hitting hard 16 v T at a high HL true count that would also be good camouflage play. The rules is hit hard 16 v T if HL + (1/2)*(AA78mTc) + 3*(5m6c) < 0. So if 5m6c is negative enough, player would hit hard 16 v T even with a high HL count.
So with HL with AA78mTc and 5m6c you have camouflage plays of insurance, hard 12 v 2, 3, 4, 5 and 6 and hard 16 v 7, 8, 9 and T that would confuse casino if they are using HL and observing your LARGE bets and comparing to what you should be doing with just the HL count.
You are only interested in camouflage of your large bets which is what the casinos are looking at so as a stated earlier, your question should be what percentage of your large bets do the camouflage plays come in at to help you stay below the casino radar. As I stated earlier CC(AA78mTc, HL) = 20% and also CC(5m6c, HL) = 0 so these two counts are basically independent of HL and can have values that are not connected to HL and so can make these camouflage plays worthwhile. Also SD(AA78mTc) = SD(HL) so AA78mTc is as variable as HL is available so AA78mTc varies and a lot and is almost independent of HL making for more camouflage plays. Also 5m6c is TOTALLY independent of HL so it can go all over the place. And SD(5m6c) is about half SD(HL) so it still varies about half as much as HL varies.
But without simulations I really cannot say what percentage of your large bets (say what percentage of your bests when HL >= 3*dr) that these camouflage plays would provide you cover. Because there is insurance, hard 12, v 2, 3, 4, 5 and 6, and hard 16 v 7, 8, 9, T potential camouflage plays and because SD(AA78mTc) and SD(5m6c) are still large and CC of AA78mTc and 5m6c with HL is small, that there should be a significant percentage of times that these camouflage plays would come into play and if they come into play and the casino notices these "mistakes" it could buy you extra time of have the casino mark as a non-counter or a mediocre counter and so you can be ignored. I would think that all you need is one or two of these "mistakes" with your large bet out of the casino to label you as a non-threat.
Again, I do not know how to do simulations so I cannot answer your question directly, only indirectly though inferences of CC and SD and number of plays where camouflage would apply to.
Last edited by bjanalyst; 01-25-2019 at 10:43 AM.
Advantages of KO + k1*(AA89mTc) + k2*(5m7c) (which was not analyzed) over HO2 with ASC.
1. KO is unbalanced with a pivot at a true count of 4. I will use HL instead of HO2 here to compare as HL and KO have approximately the same SD and so their true counts are comparable. But the same logic applies to HO2. So HL is balanced with a pivot of a true count of zero. I showed in detail in a previous exhibit comparing HL and KO true count calculations at various true counts that the farther from the pivot the more sensitive true count calculations are to errors in estimating decks remaining. At the pivot itself, the true count calculation is exact. So if HL is at its pivot of a true count of zero then HL = 0 and HL true count = 0 which is exact. If KO is at its pivot of a true count of four then KO = 4*n where n = number of decks and KO true count = 4 which is exact. It true count (tc) = 2 then HL is two true count points away from its pivot of tc = 0 and KO is two true count points away from its pivot of tc = 4 so both HL and KO are equally sensitive to errors in estimating decks remaining. If tc = 3 then HL is three true count points away from its pivot of tc = 0 but KO is only one true count point away from its pivot of tc = 4. Thus KO is only (1/3)rd as sensitive to errors in estimating decks remaining as HL is. Using similar reasoning at tc = 4, KO is exact, at tc = 5 KO is only (1/5)th as sensitive in errors in estimating decks remaining as KO is and at tc = 6 KO is only (2/6) = (1/3)rd as sensitive to errors in estimating decks played at HL is. So if tc is between 3 and 6, for example, then KO true count calculations are at least three times more accurate than HL true count calculations. Thus for KO decks remaining being estimated to the nearest full deck is equivalent to HL having to estimate decks remaining to the nearest third of a deck or better. Thus KO true count calculations are more accurate than HL true count calculations for true counts greater than two.
2. Side count of Aces is Adef = Ap – 4*dp where Adef = deficiency of Aces remaining in the shoe, Ap = Aces played and dp = decks played. Adef is APPROXIMTAE is it depends on an ESTIMATE of decks played. If your estimate of dp is off by even one-half a deck then Adef is off by 2. Plus/minus side counts calculations, AA89mTc and 5m7c, on the other hand are EXACT as they do not depend on decks played.
3. I have shown in previous exhibits that it is very easy to keep one plus minus side count and still easy to keep two plus minus side counts using chips, if necessary, to help with keeping the two side counts.
4. There are many camouflage plays that the KO with its side counts helps with, some of which I had addressed in earlier posts. If the casino is using HL and the I18 to determine if you are a counter or not these camouflage plays, which are actually the correct plays to make, will appear to the casino as if you are making mistakes and may buy you some time as they will label you as a mediocre counter.
5. I attached in previous posts the Betting Correlation (BC) of KO + (1/2)*(5m7c) and HO2 – 2*(Adef) for the S17, DAS, no LS game. KO + (1/2)*(5m7c) came in at 99.1% and HO2 – 2*(Adef) came in at 98.5%. Thus the KO system beats the HO2 system by 0.6% for betting. For the shoe game betting is very important. HO2 with ASC beat HL with AA78mTc and 5m6c because BC of HL + (1/3)*(5m6c) for S17, DAS, no LS was 97.4% and BC of HO2 – 2*(Adef) was 98.5% which was 1.1% better than this HL systems. By examining the CC of I18, it was shown that HL with AA78mTc and 5m6c beat the HO2 with ASC in 14 of the I18, tied once and lost 3 times. The losses were some and some of the wins were very large. Thus for playing strategy, HL with AA78mTc and 5m6c resoundingly beat HO2 with ASC. However for the shoe game, betting is very, very important. And with HO2 system being 1.1% better for betting than HL still put HO2 over the top.
6. KO system does not have the betting problem that HL with AA78mTc and 5m6c had. Comparing KO between with HO2 betting the tables are not turned on the HO2. KO + (1/2)*(5m7c) has a BC for S17, DAS, LS game of 99.1% and HO2 – 2*(Adef) has a BC of 98.5% as described earlier. Thus instead of the HL with AA78mTc and 5m6c being a 1.1% BC underdog to the HO2 with ASC, the HO2 with ASC is now the underdog being 0.6% below the KO system.
7. I will attach to this analysis the CC of the I18 comparing KO with AA89mTc and 5m7c to HO2 with ASC and HL with AA78mTc and 5m6c. The main deficiency of using 5m7c instead of 5m6c for the KO system is for hard 16 v T hit/stand decision which is ranked the #2 decision of the I18. Using 5m6c gives a CC of over 90% whereas using 5m7c the CC is around 70%. That is the only main playing strategy disadvantage of using the 5m7c instead of 5m6c. But all of the other advantage of the 5m7c make up for this 16 v T disadvantage. Looking at the I18 chart comparing KO + k1*(5m7c) + k2*(AA89mTc) compared to HO2 + k*(Adef) it can be seen that the KO system beats that HO2 with ASC in 11 of the I18 with CC advantage as high as 17.5%. There is one tie in the I18 and six losses with two of those six losses being less than 1%, three of the losses in the 2% range and the largest loss being 7. 4%. From this analysis it is clear that KO with AA89mTc and 5m7c beats HO2 with ASC for playing efficiency.
8. So to summarize:
a. KO system gives much more accurate true counts than HO2 when KO true count is larger than 2 which is where accurate true count calculations are most important.
b. AA89mTc and 5m7c side counts are EXACT whereas Adef is APPROXIMATE because it depends on an ESTIMATE of decks played whereas AA89mTc and 5m7c is independent of decks played.
c. The KO system has many camouflage plays which could confuse the casino if they are using the HL with I18 to track your play to see if you are a counter.
d. KO + (1/2)*(5m7c) has a BC of 99.1% whereas HO2 – 2*(Adef) has a BC for 98.5%. Therefore this KO systems beats that HO2 with ASC by 0.6% for betting and betting is very important for the shoe game.
e. An analysis of the I18 as shown above shows that the KO system beats the HO2 with ASC for playing strategy also.
f. Although no simulations of KO with AA89mTc and 5m7c were done, since the KO system beats HO2 with ASC for both betting and playing strategy, I would conclude that the simulations would show that KO with AA89mTc and 5m7c is more powerful than HO2 with ASC. And this ignores the benefits I mentioned in I, 2, and 3 above, namely, more accurate true count for true counts greater than 2, exact side count calculations as opposed to approximate ASC calculations, extremely easy to keep plus minus side counts and camouflage plays that this KO system has to throw off the casino using HL with I18 to see if you are counting or not.
Attachment 3427
Attachment 3428
Attachment 3429
Below are the results of the simulations.
All the simulations were made using the same sim parameters as specified in BJA3 Chapter 10, Page 211, except that:
- My software was used to run the simulations (BJA3 used CVCX)
- The count systems were as specified in the table below (BJA3 used HiLo)
- The index sets were as specified in the table below (BJA3 used I18 + Fab4)
- Some different bet spreads were used
- Only optimal betting to the nearest dollar was considered (BJA3 also considered practical betting ramps)
- Only one game was analyzed: 5/6 S17 DAS DOA SPL3 noRSA (BJA3 analyzed many rule sets)
In the table below:
- The full HiLo 1994 indices as obtained from CVData were used. However, since they were created using truncation for true count calculations and these sims used flooring, all negative indices were reduced by 1.
- The full HiOpt II + ASC indices as obtained from CVData were used.
- HiLo+AA78mTc 6 refers to the original six playing decisions proposed by bjanalyst using the side count of A=2, 7=1, 8= 1, T=-1 which were insurance and 12 vs 2-6
- HiLo+AA78mTc 12 refers to the addition of 6 playing decisions: 9 vs 2; 11 vs A; 13 vs 2; 13 vs 3; 15 vs T; 16 vs T
- HiLo+AA78mTc 26 refers to the addition of 14 playing decisions: 14 vs T; A,3 vs 4; A,4 vs 3; A,4 vs 4; A,5 vs 3; A,6 vs 2; A,7 vs 2; A,7 vs A; A,8 vs 3; A,8 vs 4; A,8 vs 6; 2,2 vs 8; 3,3 vs 8; 4,4 vs 4
- HiLo+AA78MTc+5m6c refers to the addition of 5 playing decisions using an additional side count of 5=1, 6=-1: 16 vs 7; 16 vs 8; 16 vs 9; 16 vs T; T,T vs 6
- The "Unlimited" bet spread is full optimal betting with no restrictions on spread nor on the maximum and minimum bets. Is it equivalent to betting optimally while wonging in place at the table. While not practical for actual play, it does represent the upper limit on the SCORE which can be achieved using each system.
Executive Summary
- While the addition of the side counts and the additional indices at each stage showed a small improvement over using HiLo alone, the stated goal of outperforming HiOpt II + ASC was not achieved. The additional side counts and indices managed to close about one half of the gap between HiLo and HiOpt II + ASC.
- The initial introduction of AA78mTc 6 showed the greatest improvement with the addition of the additional indices and side count showing only incremental gains.
The Results
Code:Scenario System Source SCORE Improvement ---------------------------------------------------------------------------- Play-All 1-8 HiLo 1994 Gronbog 24.30 HiLo+AA78mTc 6 Gronbog 27.18 11.85% HiLo+AA78mTc 12 Gronbog 27.89 2.61% HiLo+AA78mTc 26 Gronbog 28.35 1.65% HiLo+AA78mTc+5m6c Gronbog 29.53 4.16% HiOpt II + ASC Gronbog 34.49 16.80% Play-All 1-10 HiLo 1994 Gronbog 29.27 HiLo+AA78mTc 6 Gronbog 32.33 10.45% HiLo+AA78mTc 12 Gronbog 33.07 2.29% HiLo+AA78mTc 26 Gronbog 33.54 1.42% HiLo+AA78mTc+5m6c Gronbog 34.83 3.85% HiOpt II + ASC Gronbog 40.03 14.93% Play-All 1-12 HiLo 1994 Gronbog 33.04 HiLo+AA78mTc 6 Gronbog 36.21 9.59% HiLo+AA78mTc 12 Gronbog 36.95 2.04% HiLo+AA78mTc 26 Gronbog 37.44 1.33% HiLo+AA78mTc+5m6c Gronbog 38.80 3.63% HiOpt II + ASC Gronbog 44.19 13.89% Back-Count 1-1 HiLo 1994 Gronbog 46.86 HiLo+AA78mTc 6 Gronbog 49.64 5.93% HiLo+AA78mTc 12 Gronbog 50.03 0.79% HiLo+AA78mTc 26 Gronbog 50.43 0.80% HiLo+AA78mTc+5m6c Gronbog 51.67 2.46% HiOpt II + ASC Gronbog 55.70 7.80% Back-Count 1-2 HiLo 1994 Gronbog 55.56 HiLo+AA78mTc 6 Gronbog 58.86 5.94% HiLo+AA78mTc 12 Gronbog 59.45 1.00% HiLo+AA78mTc 26 Gronbog 60.06 1.03% HiLo+AA78mTc+5m6c Gronbog 61.44 2.30% HiOpt II + ASC Gronbog 66.77 8.68% Back-Count 1-4 HiLo 1994 Gronbog 61.24 HiLo+AA78mTc 6 Gronbog 64.49 5.31% HiLo+AA78mTc 12 Gronbog 65.01 0.81% HiLo+AA78mTc 26 Gronbog 65.60 0.91% HiLo+AA78mTc+5m6c Gronbog 66.98 2.10% HiOpt II + ASC Gronbog 72.31 7.96% Back-Count 1-8 HiLo 1994 Gronbog 62.91 HiLo+AA78mTc 6 Gronbog 65.96 4.85% HiLo+AA78mTc 12 Gronbog 66.43 0.71% HiLo+AA78mTc 26 Gronbog 67.00 0.86% HiLo+AA78mTc+5m6c Gronbog 68.32 1.97% HiOpt II + ASC Gronbog 74.35 8.83% Back-Count 1-12 HiLo 1994 Gronbog 63.23 HiLo+AA78mTc 6 Gronbog 66.48 5.14% HiLo+AA78mTc 12 Gronbog 67.05 0.86% HiLo+AA78mTc 26 Gronbog 67.68 0.94% HiLo+AA78mTc+5m6c Gronbog 69.11 2.11% HiOpt II + ASC Gronbog 75.05 8.59% Unrestricted HiLo 1994 Gronbog 63.87 HiLo+AA78mTc 6 Gronbog 67.01 4.92% HiLo+AA78mTc 12 Gronbog 67.55 0.81% HiLo+AA78mTc 26 Gronbog 68.16 0.90% HiLo+AA78mTc+5m6c Gronbog 69.54 2.02% HiOpt II + ASC Gronbog 75.22 8.17%
I want to thank Gronbog for doing an EXCELLENT job. He is a true professional and I appreciate his great work.
I just would like to clarify one item. It appears that even when 5m6c was added to HL with AA78mTc that there was no adjustment made to the HL for betting. If that is the case then only the HL alone was used for all simulations and there was no betting adjustment for the 5m6c. As I had shown in Betting Correlation (BC) for S17, DAS, no LS, the BC of HL is 96.48%, the BC of HL + (1/3)*(5m6c) is 97.38% and the BC of HO2 - 2*(Adef) is 98.45%. So it appears that in these simulations there was no adjustment to the HL for betting so for betting purposes the HL was at a 98.45% - 96.48% or 1.97% disadvantage.
It is to be noted that with each additional set of strategy changes using more situations with AA78mTc and then adding 5m6c there was an improvement in simulation results. But all improvements relied only on playing strategy changes and I believe that there was no attempt to improve HL betting in any of these simulations. I believe that for the shoe game we have hit the limit with just playing strategy change gains. As I had shown earlier, for CC with the I18, HL with AA78mTc and 5m6c won 14 times, tied once and lost 3 times. So I would conclude that HL with AA78mTc and 5m6c wins playing strategy efficiency over HO2.
But the unaltered HL for betting losses terribly to the HO2 - 2*(Adef) for betting. For the shoe game betting is very important. So the playing strategy gain of HL with AA78mTc and 5m6c over the HO2 with ASC was not enough to overcome the greater betting efficiency of HO2 - 2*(Adef).
If betting had used HL + (1/3)*(5m6c) the 1.97% disadvantage below the HO2 - 2*(Adef) would have been reduce to a 1.07% below the HO2 - 2*(Adef) disadvantage. But it does not appear that the simulations had any adjustments for HL betting and just HL alone was used for betting.
I want to thank Gronbog again for his great simulation results. It sheds a lot of light on plus/minus side counts and shows that plus/minus side counts work great. And I believe that if betting efficiency of HL were improved by using HL+ (1/3)*(5m6c) for betting the gap would have been closed even further.
I think we all need to thank Gronbog for this GREAT job. I truly appreciate his valuable work!
Last edited by bjanalyst; 01-25-2019 at 01:28 PM.
Another count debate.
Amazing........
Anyone surprised with the results? So, I can train and learn a new count that is harder than most humans can not accomplish only to learn it falls short of HIOptII ASC, which we have all know about forever.
Appreciate bjanalyst's enthusiasm and without his adept math skills I doubt he would have captured so much attention. I continue to suggest there are more lucrative improvements in the area of betting and longevity than trying to squeek out another bit from playing. There was a reason that the Illustrious 18/Fab 4 were created and it had to do with simplicity of play and value for effort. Those concepts should be applied to this adventure.
Nice work Gronbog!
Luck is nothing more than probability taken personally!
It is the tag values. The problem is that he is applying those side counts to a level one count. If bjanalyst were to do it to a level 2 or level 3 count system the results would be even closer and maybe even outperform Hi-OPT II with ASC.
Knowing that your intention was to attack six deck S17, DAS, no LS games, why did you only attempt to improve playing and not betting? Shouldn't you improve betting for six deck S17, DAS, no LS games and playing for single and double deck games?
You are correct. Betting is more important for the shoe game and playing strategy is more important for the two deck game. I covered many, many differnt combinations in my 4th book, High Low with plus minut side counts, much more than I can ever go over in these posts.
The HL is not my count of choice for the shoe game. I chose the HL because there are canned sim programs that use the HL and so the changes needed to this programs would be miminal. My count of choise is the KO with AA89mTc and 5m7c for ths shoe game. And KO + (1/2)*(5m7c) has a BC of 99%. In the next post I will include my full charts (to supplement the I18) of the KO with AA89mTc and 5m7c.
I would like to emphasize that my Excel program I created to calculate indices and values of k for side counts has now been proven to work through simulations. Gronbog added the indices and values of "k" I have him from my program into his simulations and each time he added more AA78mTc or 5m6c improvements to the HL the sims showed improvements. In addition when I first made my program in 2011 I had ETFAN review it who was the mathematician for Arnold Synder. ETFAN was not familiar with my LSL (Least Square Line) technique but he did know Griffin's PD (Proportional Deflection) which he taught me and which I also programmed into my spreadsheet. The results showed that the LSL and PD both gave identical results. In addition my program gave the HL indices that agreed with published indices. And not simulations show that the calculated indices and values of "k" lead to improvements. And the psrc (playing strategy running count) = HL + k1*(AA78mTc) + k2*(5m6c) formulas also make logical sense. So everything falls in place. My calculations are correct.
For the two deck game I think balanced counts should be used because you play all hands and the true counts go all over the place and would not be outside a table of critical running count extremely often. So analyzed the following counts in my 4th book for the two deck game: Note that what I call HL2 = High-Low 2 is a level two version of the HL using halves -- take the HL and decrease the tag value of the 2 from +1 to +1/2 and increase the tag value of 7 from zero to +1/2 so that the count is still balanced. So HL2 has 2 and 7 as +1/2, 3, 4, 5, 6 as +1, 8, 9 as zero and T and Ace as -1. So there is hat I analyzed:
1 One side count HL with Am6c. brc = HL has betting efficiency = 96.5%. 2 One side count: Before adding a 2nd side count, switch from HL to HL2 HL2 with Am6c where HL2 is the HL with 2's and 7's counted at one-half. Use indices and "k" values of HL, Am6c for HL2, Am6c. Actual infinite deck indices and values ok "k" for HL2 with Am6c have been calculated and are included in this chapter and should be used instead of approximations. brc = HL2 has betting efficiency = 97.6%. 3 Two side counts Add 5m9c to HL2 with Am6c. Use indices and "k1" and "k2" values of HL, Am6c, 5m9c for HL2, Am6c, 5m9c. Actual infinite deck indices and values of "k1" and "k2" for HL2 with Am6c and 5m9c have been calculated and are included in this chapter and should be used instead of approximations. brc = HL2 + ½*(5m9c) is Wong's Halves with betting efficiency = 99.3%. 4 Alternate second side count Instead of switching from HL to HL2, keep the HL with Am6c and add 7m9c. For the one and two deck game, hit/stand on hard 14 v T is an important decision. Adding 7m9c increases the CC of this decision from 41% to 78%. 7m9c also helps with betting where brc = HL + ½*(7m9c) with betting efficiency of 98.1%.
Attached is my more complete analysis of KO with AA89mTc and 5m7c.
I showed in previous posts that keeping two plus/minus side counts is easy. And using them is easy. You are just mutiply and addding integers and comparing the rsuls to a thrid integer, i.e. make a playing strategy change when psrc = KO + k1*(AA89mTc) + k2(5mc) >= crc(Idx) where Idx = index for that particular strategy change and k1 annd k2 are the values for that startegy change that maximize the CC between the tag valuse of the derived count, psrc, and the EoR from Schelsinger's BJA 3rd edition.
My analysis shows that BC for KO + (1/2)*(5m7c) beats HO2 - 2*(Adef) and that KO, AA89mTc and 5m7c CC beat HO2 w ASC for the majority of the I18 situations. In the attached charts you will see that there are even more strategy changes that could be camouflage plays when 5m7c is used with KO and AA89mTc than when 5m6c is used as the 2nd side count.
I emphasized KO, AA89mTc and 5m7c as my combination of counts of choice for the shoe game in my 3rd and 4th books. I did not want Gronbog to analyze those counts to begin with because that amount of changes to the sim programs would have been excess and errors could have been made. Much easier to prove that my technique was correct by using the HL with AA78mTc and 5m6c which he did. So you should be confident that my attached charts are no correct also. And based on the fact that KO with AA89mTc and 5m7c had a higher Betting Efficiency and has higher CC for the vast majority of playing decisions tan HO2 with ASC does and so has a higher playing efficiency. then the appropriate conclusion is that the KO with AA89mTc and 5m7c beats that HO2 with ASC and also have a lot of camouflage plays.
Betting Efficiency KO with 5m7c (1).jpg
Betting Efficiency KO with 5m7c (2).jpg
I18 KO AA89mTc 5m7v vs HO2 ASC.jpg
table KO with AA89mTc and 5m7c.jpg
chart KO with AA89mTc and 5m7c.jpg
Posting Permissions
-*-
Bookmarks