Experiments of this weekend...

[mr.ore]

Well, during this weekend, after quite a long time, I felt like playing roulette and continue to develop some systems.           I'm trying to "master" even chances betting, and came with interesting idea:

Let's make table, which tells us, how much to bet on EC according to our session balance and loses in a row:



loses in row :  0   1   2   3   4   5 .          .          .         
balance:

+512:          500  64  .          .          .         
  .           
  .                                    .          .          .          .         
  .                 
 +1               2     1   1    1   1
  0                 1    1   1    1   1
  -1               2     1   1    1    1
  .         
  .                                 .          .          .          .         
  .         
-512           500  128

There are some values, which I am now trying to find, which are best for a given spread of bets (in this example min.           bet is 1 unit and max.           bet is 500 units = common spread in internet RNG casinos).           Values in that table above are just for example, I am developing formula to compute them, it actually combines idea of positive and negative progression and lowering bet after lost.           I came with the idea by playing Oscar's grind, many times happened, that just before breaking even it failed into big hole, because bet was unnecessary big.           This table tries to prevent it by defining bet for each balance.           There is also posibillity for positive progression, so if we get lucky before breaking even, we break even with it.         

My values for table in this test are given by this:

for (int I=0; I<m_MaxBet; I++) {
 for (int j=0; j<64; j++) {
    m_Table[512-I][j] = I * 0.          5  + 1;
    m_Table[512+I][j] = I * 1.          0 / (j*0.          25+1) + 1;
 }
}

Well, practically, there won't be more than 37/18*20 = 41 loses in a row, that 64 is just because of alignment (or more because I like powers of 2   )

The money management is like this: my target is 8 units, and stop loss is 32 units, and after lost session, I increase unit by one, after win I reset it to one if overall balance is positive, and decrease by one (a la Contre D'Alembert) if balance is negative.           Any time we got to positive or negative balance, I set unit to one and reset.     

Note&edit:  I found a mistake in my code.     .     .     .        so experiemnt worked like this:

real bet = table[-Unit][MM lost sessions in a row]*table[balance][loses in a row], and real bet is truncated to 1.     .     500 interval after this.      -Unit because we are using Contre D'Alembert as MM.      So it is more bold play than I expected, there were just too much mess in my code, so I overlooked it.     .     .     

Edit2&: Another error in code, after lost sesion, I increased unit instead of decrease, and only if unit was biger than one, ie.  after MM's session won. . .  this is crap.  So the results are for this type of MM, Contre D'Alembert do not work as MM for my simulations, but I may have done another mistake.  I'm just now having losing streak in coding mistakes 

It is loser of course, but with those experiments with bankroll managment, I understood, how could it be even possible to make system which can play for thousands of spins.           Bankroll managment is definitely important, same rules for bet, do not bet too much if we are losing, do not overkill it and rather wait for a lucky series which will help us.       

[mr.ore]

One of experiments ended up promising:

[mr.ore]

Another failed badly:

[mr.ore]

But sometimes it can recover from huge lose, and max.  bet is still just 500 units:

[mr.ore]

It can win a lot of units, but fails in the end.   Also, there are times where it could have used existing win to gain much more, but it was just waiting for a series, there I must came with idea how to better capitalize on such places, it have to win as fast as possible, for it will fail in the end. 

[mr.ore]

-