1. Welcome to the #1 Gambling Community with the best minds across the entire gambling spectrum. REGISTER NOW!
  2. Join our $5,000 Cash Giveaway!

    Win Cash by Posting and Inviting New Members!
  3. Have a gambling question?

    Post it here and our gambling experts will answer it!
    Dismiss Notice
  4. Discussions in this section are assumed to be EV- as they are outside of the Advantage Play section. For EV+ discussions, please visit the Advantage Play sub-forum.
    Dismiss Notice

Baccarat Neural Strategy

Discussion in 'Baccarat Forum' started by Jimske, Jul 29, 2019.

  1. soxfan

    soxfan Active Member

    Joined:
    Dec 29, 2016
    Likes:
    82
    Location:
    FrozenTundra
    This neural thingy look like a modify version of the Barstow style, hey hey.
     
  2. gizmotron

    gizmotron Well-Known Member Founding Member

    Joined:
    Jan 1, 2015
    Likes:
    422
    Occupation:
    Retired
    Location:
    The West Coast of USA, RV'ing
    OK, so I slapped together a quick power test where the bet selection is only on Red to win on a single zero wheel. As we all know experts on gambling all know that the next bet is just a guess anyway. So here are the results of just betting on red using the super duper 7 steps 5, 8, 13, 20, 35, 50, 75, that all add up to 206 if it loses a 7th step.

    Code:
       W -- 5 -- 5
    L    -- 0 -- 5
    L    -- -8 -- 8
    L    -- -21 -- 13
    L    -- -41 -- 20
       W -- -6 -- 35
       W -- 14 -- 20
       W -- 27 -- 13
       W -- 35 -- 8
       W -- 40 -- 5
       W -- 45 -- 5
       W -- 50 -- 5
       W -- 55 -- 5
       W -- 60 -- 5
       W -- 65 -- 5
       W -- 70 -- 5
    L    -- 65 -- 5
    L    -- 57 -- 8
    L    -- 44 -- 13
       W -- 64 -- 20
       W -- 77 -- 13
       W -- 85 -- 8
       W -- 90 -- 5
       W -- 95 -- 5
       W -- 100 -- 5
       W -- 105 -- 5
       W -- 110 -- 5
       W -- 115 -- 5
    L    -- 110 -- 5
       W -- 118 -- 8
    L    -- 113 -- 5
       W -- 121 -- 8
    L    -- 116 -- 5
       W -- 124 -- 8
    L    -- 119 -- 5
    L    -- 111 -- 8
       W -- 124 -- 13
       W -- 132 -- 8
    L    -- 127 -- 5
    L    -- 119 -- 8
       W -- 132 -- 13
       W -- 140 -- 8
    L    -- 135 -- 5
       W -- 143 -- 8
    L    -- 138 -- 5
    L    -- 130 -- 8
    L    -- 117 -- 13
       W -- 137 -- 20
    L    -- 124 -- 13
       W -- 144 -- 20
    L    -- 131 -- 13
    L    -- 111 -- 20
       W -- 146 -- 35
       W -- 166 -- 20
       W -- 179 -- 13
       W -- 187 -- 8
       W -- 192 -- 5
    L    -- 187 -- 5
    L    -- 179 -- 8
       W -- 192 -- 13
       W -- 200 -- 8
       W -- 205 -- 5
       W -- 210 -- 5
       W -- 215 -- 5
       W -- 220 -- 5
    L    -- 215 -- 5
    L    -- 207 -- 8
       W -- 220 -- 13
    L    -- 212 -- 8
       W -- 225 -- 13
    L    -- 217 -- 8
       W -- 230 -- 13
       W -- 238 -- 8
       W -- 243 -- 5
    L    -- 238 -- 5
       W -- 246 -- 8
    L    -- 241 -- 5
       W -- 249 -- 8
    L    -- 244 -- 5
    L    -- 236 -- 8
       W -- 249 -- 13
       W -- 257 -- 8
    L    -- 252 -- 5
    L    -- 244 -- 8
    L    -- 231 -- 13
       W -- 251 -- 20
       W -- 264 -- 13
    L    -- 256 -- 8
       W -- 269 -- 13
       W -- 277 -- 8
       W -- 282 -- 5
       W -- 287 -- 5
    L    -- 282 -- 5
       W -- 290 -- 8
       W -- 295 -- 5
       W -- 300 -- 5
    L    -- 295 -- 5
    L    -- 287 -- 8
    L    -- 274 -- 13
       W -- 294 -- 20
    
    and the source code to run it is:

    Code:
    on mouseUp
       put "1,3,5,7,9,12,14,16,18,19,21,23,25,27,30,32,34,36," into rd
       --put "2,4,6,8,10,11,13,15,17,20,22,24,26,28,29,31,33,35," into blk
       put "5,8,13,20,35,50,75" into hStepsList
       put 0 into hTot
       put ""  into field "mainText"
       lock screen
       put 1 into hStep
       repeat with i = 1 to 100
          put random(37) into hSpin
          put item hStep of hStepsList into thisStepNow
          if hSpin is among the items of rd then
             put "   W -- " after field "mainText"
             add  thisStepNow to hTot
              put hTot & " -- " after field "mainText"
             put thisStepNow & return after field "mainText"
             if hStep > 1 then
                if hStep < 8 then
                   subtract 1 from hStep
                end if
             end if
          else
             put "L    -- " after field "mainText"
             subtract  thisStepNow from hTot
             put hTot & " -- " after field "mainText"
             put thisStepNow & return after field "mainText"
             if hStep < 7 then
                add 1 to hStep
             end if
          end if
       end repeat
       unlock screen
    end mouseUp
    
    This is just one 100 spin test. I can now go on to power test it with thousands of tries.
     
  3. mansi19896

    mansi19896 Member

    Joined:
    Oct 17, 2018
    Likes:
    11
    Location:
    estonia
    isnt it better to wait 4 losses and just bet 1-3-7 progression. will it work?
     
  4. Jimske

    Jimske Active Member Founding Member

    Joined:
    Dec 30, 2014
    Likes:
    117
    5 -8--13-20-35-50-75-100 = 8 bets.
     
  5. Jimske

    Jimske Active Member Founding Member

    Joined:
    Dec 30, 2014
    Likes:
    117
    okay but you said that you're going to test the neural strategy. If you don't want to that's fine but why don't you just say that and then move to another thread with whatever test you want to test. Why jump on this thread?

    It seems that you always do this. You always try to change and talk about your own roulette stuff.
     
    Last edited: Aug 1, 2019
    Junket King and gizmotron like this.
  6. gizmotron

    gizmotron Well-Known Member Founding Member

    Joined:
    Jan 1, 2015
    Likes:
    422
    Occupation:
    Retired
    Location:
    The West Coast of USA, RV'ing
    Thanks,

    So I wrote a few power algorithms that tests thousands of spins with hundreds of sessions. I used only betting on red on a single zero wheel. I picked several different win stop points looking for the best stop win point. As you would expect they all lost money. The average loss turned out to be between 21 to 54 units per session. But that was just blind bet selection that just barges right through all the losing streaks. My experience with these things is that if you use the Neural Strategy's bet selections you will get just a slight improvement over blind random on random bet selection.

    The key to these tests is to find out if the Neural Strategy's bet selection makes much of a difference. I don't think it will because it also barges right through losing streaks. I know that the rules have it making adjustments for rule based conditional changes. But these rules don't allow for the effectiveness states. That is the missing rule. It's a pain in the ass to write an Effectiveness State rule into testing software. I did it once just to prove that I could build Artificial Intelligence, AI. I learned after doing it that my own skill at avoiding bad sequences of spins or Pass / Don't Pass bets was far more acute. So I won't put myself through all this trouble and time when I know where the weakness is in this progression and bet selection method.

    I would be interested if I were to get paid more than $500 per hour in programming time. It's just a massive headache. You must hard code hundreds of situations that are all around pattern recognition. Then these unique situations can be processed through a class library of algorithms that work as Object Oriented Programming functions. In other words it is creating computer processing library of functions to handle randomness characteristics. It's doable but it is a huge undertaking and would be best served as an Open Source project by thousands of interested programmers.

    Worst of all, I already know the final outcome. I know that the rules, as they exist now for bet selection changes, will barge right through losing streaks. So that is where adjustments need to be made. It looks pretty good for protecting your bankroll but it still loses in the long run as it is now. Because I'm still much better at reading randomness, for much more bet selection adjustments and changes, I can best serve this research by just practicing a few sessions per day using the step up and step back down progression against Effectiveness States hunting. That's like knowing when not to barge through with these progression steps or even flat betting for that matter. Still, I really like this progression because it does not return back to the base value of one after any win. It just steps back down one step at a time. So it feeds itself winners logarithmically during recovery.

    Thanks for sharing this interesting method.

    Mark
     
  7. gizmotron

    gizmotron Well-Known Member Founding Member

    Joined:
    Jan 1, 2015
    Likes:
    422
    Occupation:
    Retired
    Location:
    The West Coast of USA, RV'ing
    I had no idea that this was a stepped progression both upward and downward. It acts differently than progressions that return to the base starting point after each win. I have years experience at assessing bet selection changes as simple rules that must be followed. That experience allows me to already see where a thing is going. I know I went over your head and barged right in with my own stuff. I apologize for that. It tends to make me appear condescending. Once I saw these steps up and steps down I was intrigued. I had not researched this kind of rule based system before. I will get the hell out of your thread. Thanks again for sharing it.
     

  8. gizmotron

    gizmotron Well-Known Member Founding Member

    Joined:
    Jan 1, 2015
    Likes:
    422
    Occupation:
    Retired
    Location:
    The West Coast of USA, RV'ing
    If anyone wonders why I still hang around gambling forums then you have just seen why. I like making a selfish pest of myself. Actually I like seeing things that I have never considered before. This thread is a huge breath of fresh air to me. I have the ability to test it and find its strong points. That comes from years of researching this kind of stuff. Anyway, thanks once again for the ideas. It people like Jimske that want to discuss ideas that make gambling forums more interesting.
     
    Last edited: Aug 1, 2019
    Junket King likes this.
  9. stephen

    stephen New Member

    Joined:
    Dec 7, 2017
    Likes:
    7
    Location:
    USA
    gizmotron,

    Thanks for simulation. The average loss per session is 21 to 54 units for single zero wheel playing red (HA 2.6%). Can you simulate for single zero wheel playing red with la partage rule (house advantage of 1.3%) which corresponds approximately to Baccarat HA of 1.15% and give the average loss per session. This will require some change to your program. I am curious to know "What is the approximate loss per session for HA of 1.3%".
     
  10. gizmotron

    gizmotron Well-Known Member Founding Member

    Joined:
    Jan 1, 2015
    Likes:
    422
    Occupation:
    Retired
    Location:
    The West Coast of USA, RV'ing
    Interesting situation. The chance to win or lose still exists. You will probably never see seven lost bets because of a streak of seven zeros. It will likely mitigate one of the seven steps per session and that must occur during a seven step loss. Still, you would get back half of whatever bet you lost on the zero hit. Let Jimske tackle this. I'm done programming this. I know it won't be enough to move it into a win column. You will only see two or three zeros on average for each 100 spins. These zero events must occur during a seven spin losing streak. It's the loss that is more important than the split. Now if there was a step rule for zero losses that would be interesting. The step increase should reflect the la partage.
     
  11. gizmotron

    gizmotron Well-Known Member Founding Member

    Joined:
    Jan 1, 2015
    Likes:
    422
    Occupation:
    Retired
    Location:
    The West Coast of USA, RV'ing
    I got 21 to 54 as the average long run loss from stop win points set at from 50 to 500. I tested everything at 50 increased intervals, stops at 50, 100, 150, etc...
     
  12. gizmotron

    gizmotron Well-Known Member Founding Member

    Joined:
    Jan 1, 2015
    Likes:
    422
    Occupation:
    Retired
    Location:
    The West Coast of USA, RV'ing
    Today's practice session using stepped progression: Steps are 1, 2, 3, 5, 8, 10, and 15, where each step is times 18 numbers straight up for a bankroll of 800. I stopped at 25% win on bankroll. Bet selection technique is reading randomness & virtual betting.

    Code:
     W | B  R | O  E | L  H | 0  6 | P | S |  --  SN  --  SP 
     nb| X    |    X | X    |    X |   |   |  --  01  --  08 
     nb| X    |    X |    X | X    |   |   |  --  02  --  26 
     nb|    X | X    | X    | X    | X |   |  --  03  --  01 
     nb|    X | X    |    X |    X |   |   |  --  04  --  19 
     nb|    X | X    | X    |    X | X | X |  --  05  --  05 
     nb| X    |    X | X    | X    |   | X |  --  06  --  10 
     nb|    X | X    |    X |    X |   |   |  --  07  --  19 
     nb| X    | X    | X    |    X | X |   |  --  08  --  17 
     nb|--------------------| X    | X | X |  --  09  --  38 
     nb| X    |    X |    X | X    | X |   |  --  10  --  28 
      -|    X | X    |    X | X    |   | X |  --  01  --  25 ( $ 0 ) Black
     X-| X    |    X |    X | X    |   |   |  --  02  --  26 ( $ 0 ) Black
     X |    X |    X |    X |    X |   |   |  --  03  --  34 ( $ 18 ) Red
     X |    X |    X |    X |    X |   |   |  --  04  --  34 ( $ 34 ) Six's
     X |    X | X    |    X |    X |   | X |  --  05  --  21 ( $ 50 ) Not Primes
       | X    | X    | X    |    X |   | X |  --  06  --  15 ( $ 30 ) Not Specials
       | X    | X    |    X |    X |   | X |  --  07  --  33 ( $ -10 ) Not Specials
     X | X    | X    |    X | X    |   |   |  --  08  --  29 ( $ 38 ) Not Specials
       | X    | X    | X    | X    |   | X |  --  09  --  13 ( $ -2 ) Not Specials
       |    X |    X | X    |    X | X | X |  --  10  --  12 ( $ -62 ) Not Specials
     X |    X |    X |    X | X    |   |   |  --  11  --  36 ( $ 18 ) Not Specials
     X | X    |    X |    X | X    |   |   |  --  12  --  26 ( $ 66 ) Not Specials
       |    X | X    |    X | X    |   | X |  --  13  --  23 ( $ 26 ) Six's
       |    X |    X |    X | X    |   |   |  --  14  --  36 ( $ -34 ) Six's
     X |    X | X    |    X |    X |   |   |  --  15  --  19 ( $ 46 ) Six's
     X |    X | X    |    X | X    | X |   |  --  16  --  27 ( $ 100 ) High
       | X    |    X |    X |    X |   | X |  --  17  --  22 ( $ 64 ) Red
     X |    X | X    |    X |    X |   |   |  --  18  --  19 ( $ 118 ) Red
     X |    X | X    |    X | X    |   | X |  --  19  --  25 ( $ 154 ) Red
       | X    | X    |    X | X    |   | X |  --  20  --  35 ( $ 136 ) Red
     X |    X | X    | X    | X    |   |   |  --  21  --  09 ( $ 172 ) Red
       | X    | X    |    X |    X | X | X |  --  22  --  31 ( $ 154 ) Red
       | X    |    X | X    | X    |   | X |  --  23  --  10 ( $ 118 ) Red
     X |    X | X    | X    | X    | X |   |  --  24  --  01 ( $ 154 ) Red
     X |    X | X    |    X |    X |   | X |  --  25  --  21 ( $ 208 ) Red
    
    Here is the graph for this session:

    graph44.png
     
  13. gizmotron

    gizmotron Well-Known Member Founding Member

    Joined:
    Jan 1, 2015
    Likes:
    422
    Occupation:
    Retired
    Location:
    The West Coast of USA, RV'ing
    OOPS! Sorry Jimske, I thought I was in my own thread. Crap. This won't happen again. Senile

    Just look at that domination in the not primes. I missed that.
     

Share This Page