Baccarat Neural Strategy

This neural thingy look like a modify version of the Barstow style, hey hey.

 

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.

 

isnt it better to wait 4 losses and just bet 1-3-7 progression. will it work?

 

5 -8--13-20-35-50-75-100 = 8 bets.

 

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.

 

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

 

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.

 

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.

 

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%".

 

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.

 

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...

 

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:

 

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.

 

The "Neural Progression" seems like a more intelligent version of the Fibonacci. Would work nice with a Bet Selection hit rate close to 50% or better that keeps us from busting out over the top. Thanks for sharing. I'll practice with it and read that 120 page book you shared. Thanks Jimske.

 

Interesting thread, shame it has apparently died. I'mm no computer wizard as some of these posters, but I have noticed none of these wizards have mentioned anything about a target point as described. Unfortunately, that death spiral of seven loses is real and unavoidable. My wife says that's why it's called gambling, and there is no sure thing except taxes and death.

 

7 LIAR should not be a deal breaker. I've lost 7 IAR a few times in the last several years. Each time I recovered.

 

I never said it was a dealbreaker, only that it is inevitable. You may reach a target several times however, if you continue to play the death spiral is on the way. Hopefully you have reached your target enough btimes to surive.

 

Following up on my results from The Neural Strategy, here are my results. Played 34 sets, (sets being either to reach my target $ amount and begin over or crash and burn) I have won 27 sets and lost 7. Not a dealbreaker, but close. However, with this kind of track record, it takes gonads of steel to continue with a 20.5% loss record. My loses were never two in a row, but pretty much scattered throughout my play. Consequently, in overall long play, you should average somewhere between 15 % and 25 % loses. An enormous amount of time invested for very little money. My advice is not to give up your day job. I'mm not saying 34 sets is enough to establish a track record. Who knows, I could win the next twenty or more in a row. However, it's unlikely.

Maybe Jimske I am misinterpreting your post, however according to what you posted, to have lost only a few sets in several years, you either played very, very little or you are not being truthful. I'mm not saying it's not possible to recover, but it is definitely a very long shot. I'mm not trying to pick a fight with you, just pointing out some facts. Furthermore, I do appreciate your posts and would like to establish a meaningful dialog with you.

 

Beam me up, Scotty. There is no intelligent life here.

 

Nuff said .

ND