Since youve decided to be judge, jury and executioner, you might as well give the order to fire.Originally Posted by seriousplayer
https://youtu.be/_N2o6MAxecA
Since youve decided to be judge, jury and executioner, you might as well give the order to fire.
https://youtu.be/_N2o6MAxecA
I have to admit that you were correct about average CC and sim results. You have to analyze each CC for each situation individually as an average, especially a straight average, can be misleading.
Attached is a chart for each of the 118 for HL + k1*(AA78mTc) + k2*(5m6c) with HO2 with ASC SORTED by CHANGE in CC.
Now the CC of each of the I18 can be directly compared.
If HL + k1*(AA78mTc) + k2*(5m6c) beat HO2 with ASC for every single I18, then the conclusion would be that HL + k1*(AA78mTc) + k2*(5m6c) beats HO2 with ASC.
The results are that HL + k1*(AA78mTc) + k2*(5m6c) beats HO2 with ASC, as measured by INDIVIDUAL CC, in 14 of the I18, ties once and losses 3 times. The maximum win is 17.2% and the maximum loss is 11.5%. Also note a resounding 16.9% CC increase over HO2 with ASC for the very important hard 16 v T hit/stand. All losses are shown in red.
Gronbog agreed to do sims of HL + k1*(AA78mTc) + k2*(5m6c) and let's see what happens.
If the simulation results turn out to favorable for HL + k18(AA78mTc) + k2*(5m6c) then I can go into more detail on how to use these side counts. With a little practice, it is really very easy to do.
Last edited by bjanalyst; 01-22-2019 at 07:23 AM.
Sorry for using acronyms that some people may not know. CC = Correlation Coefficent between the tag values of the (derived) count and the EoR (Effects of Removal).
Attached is CC for hit/stand hard 16 v T for HL + k1*(AA78mTc) + k2*(5m6c) compared to HO2 + k*(Adef). CC of HL + 0.5*(AA78mTc) + 3*(5m6c) is 94.3% and CC of HO2 + Adef is 77.4%.
So it should be clear that for the all important hard 16 v T, HL + 0.5*(AA78mTc) + 3*(5m6c) beats HO2 + Adef
And as I showed in a previous chart CC of HL + k1*(AA78mTc) + k2*(5m6c) beats CC of HO2 + k*(Adef) in 14 of the I18, ties for one and loses for 3.
Last edited by bjanalyst; 01-22-2019 at 08:45 AM.
The biggest gain from more accurate decisions is being able to bet more with an optimal spread. Try targeting matchups that reduce variance and add a lot to EV thus increasing you optimal bet more, like doubles and splits. You will get a lot more gain for your effort. Also decisions that have big bets out around the index so more accuracy will affect your largest bets. I tried to tell you that early in the thread, but you didn't want to listen to anybody. The gain will still be minor but at least you are getting the biggest impact by being able to bet more with the same RoR so you get gain on every bet made with an advantage, not just the times you get the matchup. That is where you will see more gain.
I am not sure how optimal betting works. I showed that adding (1/3)*(5m6c) to HL increases betting correlation. I will let Gronbog figure out how to input that increase in BC into optimal betting.
But I would like to comment on more doubles and splits. Double and splits increase variance so I usually wait until true count is at least one true count point higher than the double or split index - exactly at the index it is even and does not matter whether you double or split or not - so why double or split right at the index where there is virtually no extra expected value gain at the index and only increased variance?
Also when you see Gronbog's simulations, when AA78mTc was added to HL there was some improvements in doubles and splits, but the extra gains were marginal. Most of the extra gain will be obtained by targeting the I18 and increasing the CC of the I18 rather than learning extra indices and exotic doubles and splits. So that is exactly what I did here - I targeted in CC of the I18 with HL + k1*(AA78mTc) + k2*(5m6c).
Where I play there is S17, DAS, LS and LL offered. Late Surrender (LS) helps reduce risk and increasing advantage so your optimum bet can be increased. Gronbog analyzed S17, DAS, no LS. So simulations do not involved LS. Also I play LL (Lucky Ladies) which is a side bet which can have big edges which I covered earlier in this thread. Again, no simulations of side bets here. These simulations are plain, stand alone blackjack with S17, DAS and no LS.
By the way, I do not suggest the HL for the shoe game. I suggest KO. The HL sim programs are already available and HL is what most counters use, thus I analyzed HL with plus minus side counts here. My preference is for KO with plus/minus side counts. Thus my preference is KO + k1*(AA89mTc) + k2*(5m6c) as opposed to HL + k1*(AA78mTc) + k2*(5m6c) for the shoe game precisely because of increased TC accuracy at true counts >= 3 which is what you are most interested in anyhow since that is where you have your largest bets. See attached file comparing TC accuracy of HL with KO.
True Count Accuracy.jpg
Last edited by bjanalyst; 01-23-2019 at 07:50 AM.
Please look at attached files. They show comparison between traditional side counts and plus/minus side counts.
Plus/Minus sides counts are extremely easy to keep and are EXACT. HO2 with ASC is an extremely difficult count to keep and Adef is an ESTIMTATE as it relies on an estimate of decks played.
I will give more examples of how easy it is to keep HL with AA78mTc and 5m6c in a future post.
Keeping two side counts1.jpg
Keeping two side counts (2).jpg
Keeping two side counts (3).jpg
Last edited by bjanalyst; 01-23-2019 at 05:27 AM.
Bjanalyst, You studied research BRH ?
https://www.blackjackreview.com/wp/a...s/brh-systems/
"Don't Cast Your Pearls Before Swine" (Jesus)
Sigh.
You don't get it, do you?
First off: keeping HOII w/ ASC is significantly easier than keeping three independent running counts, especially since some of the tag values collide! (your 5/6 SC versus High Low)
Second: you had simulation work done that demonstrated that using CC is a poor method of determining what is "best"; comparing system CC's does nothing! What is the win rate for each system for the given example? What is the SCORE?!
Third: traditional side counts are accurate in the sense you know the total number of any given rank. singe-parameter point counts do no such thing! (caveat: the Ten's count is the only I can think of.)
Lastly: There is no one that I am aware of that strictly uses 8's for side counting with HOII. Side counting the Ace and 7 for 1D and 2D is recommended only and for 6/8D, simply use the ASC for betting.
All in all, your attempt to research your system via sim is commended! However, you attempt at pushing your system as better than HOII is bullshit.
Consider simply sticking w/ High Low and forget publishing your system! Very few (if any!) will buy it.
You never tried plus/minus side count and you are stuck with HO2 with ASC because that is all that you know. I tried the HO2 and I think it is VERY difficult and that is without any Ace side count. A level two count is much more difficult than a level one count. People tend to hate change and that is what I am encountering here - they resist change and love the status quo. So I would like to address each of your four points separately:
First off: keeping HOII w/ ASC is significantly easier than keeping three independent running counts, especially since some of the tag values collide! (your 5/6 SC versus High Low)
That is a matter of opinion. I think keeping a level two count and a separate side count of Aces is VERY difficult and also subject to errors in keeping both the primacy level two HO2 count and in the calculation of Adef = deficiency of Aces which is an estimate.. For example, Adef = Ap - 4*dp, where Ap = Aces played and dp = decks played, is subject to an estimate of decks played so assuming that you did not make any makes in keeping the ever increasing value of Ap, even if your estimate of dp is off by as little as (1/2) deck your estimate of Adef is off by 2. So (a) HO2 is a difficult level 2 primary count and is subject to errors (2) The calculation of Adef is an approximation of deficiency of Aces remaining in the shoe. You have probably made mistake of both HO2 and Adef calculations and you are not even aware of those mistakes. On the other hand, HL is an every easy level one count and the side counts, AA78mTc and 5m6c are both plus/minus count and so are exact and easy to keep which I will explain in greater detail after Gronbog has finished his latest simulations.
Second: you had simulation work done that demonstrated that using CC is a poor method of determining what is "best"; comparing system CC's does nothing! What is the win rate for each system for the given example? What is the SCORE?!
Wrong! The simulation work showed the AVERAGE of the CC is a poor method of determining the best count. Please read the posts carefully. I admit I was slopping in taking the easy way out and just saying, without any proof, that since the average CC of HL with AA78mTc for the I18 and the average CC of HO2 with ASC for the I18 were approximately equal to the counts must be approximately equal. The average hides a lot. What you need it to is look at the CC of each situation individually. I explained my error in a previous post.. In particular, look at the second most important playing strategy change which is hit/stand hard 16 v T where I gave attached files with the actual calculations. \ The CC, when analyzed for this individual situation are: HO2 76.5%, HO2 & ASC 77.4%, HL 55.8%, HL & AA78mTc 65.1% and HL & AA78mTc & 5m6c 94.3%. So when simluations where done with HL & AA78mTc for 16 v T the CC of 65.1% was 12.3% below HO2 & ASC but now adding 5m6c is 16.9% above HO2 & ASC for a swing of 29.2% in CC. Wait to see what the sim results say about this. I think this one play alone had a big influence in HL with AA78mTc gained only half way from HL to HO2 with ASC. Remember, the sims were for no LS so two card 16 v T could not be surrendered which make 16 v T even more important for the no LS game.
Third: traditional side counts are accurate in the sense you know the total number of any given rank. singe-parameter point counts do no such thing! (caveat: the Ten's count is the only I can think of.)
Why do you think you need to know the exact number of any given rank to make a great system? I showed that knowing an exact value of Adef for HO & ASC for hard 16 v T gave only a CC of 77.4% which increased HO2 CC of 76.5% by only 0.9% and fell 16.9% below HL with AA78mTc and 5m6c. So where is the benefit in knowing the exact number of a given rank in making your playing strategy decisions? You made a statement without any proof. Similar to me making the mistake to just use average CC alone to measure strength between various systems. We both need to think before we make statements without proof.
Lastly: There is no one that I am aware of that strictly uses 8's for side counting with HOII. Side counting the Ace and 7 for 1D and 2D is recommended only and for 6/8D, simply use the ASC for betting.Your attempt at pushing your system as better than HOII is bullshit.
I used this as an example as this was the example I found on the Internet. So now you admit that keeping another side count 8def is DIFFICULT with the HO2. So what I showed is that using the simple HL with AA78mTc gave results for 9 v 2 double equal to the very difficult HO2 with ASC and 8SC. And that assumes that you calculated 8def exactly which you know is only an approximation since dp is an approximation. But AA78mTc side count is exact. So you statement just disproved your own argument and proved my point. I will show after simulation results how easy it is to keep HL with AA78mTc and 5m6c .
Let's wait to see the simulation results of HL with AA78mTc and 5m6c before you draw any conclusions.
Last edited by bjanalyst; 01-23-2019 at 06:25 AM.
Posting Permissions
-*-
Bookmarks