You are totally right and we know all ththe facts too.
Martingale is the worst progression ever and does not work, there are some better progression yes but they will all fail if you are basing your strategy only in a progression because not matter how good its it will not overcome the house edge, the only way to so it is increasing your bet selection prediction.
A winning strategy need the following:
Flat bet/progression + bet selection + money management
Let me explaine you what we had done:
We had develope a custom hybrid progresion that use flat bet, positive and negative trend, which means that your balance will increase and decrease in the midle of the session.
For bet selection our algorithm increase the Accuracy of the results enought to overcome house edge.
For money management we had set 200 units as min capital and Stop Loss feature to not let you loss more than 20% of your capital in a bad session.
We had work very hard to provide you with a winning gambling system and i think that 1 week is annough time to evaluate the results, because lets be frank have you ever gamble for 1 week and had continue winning without lossing money for the whole week ?
Isnt it enought a 1 week TST? Have you ever TST a gamble system that can hold for 1 week ?
To the people not interesting, continue lossing and move on please.
Thanks
I'm not certain how your algorithm is supposed to detect a sufficient bias using a "positive or negative" trend. The next result in a dice game is not dependent on what happened in any of the previous results, unless there is some kind of flaw in the random number generator. (Unlike a card game where the cards are not shuffled, which is dependent on previous results since the same card can't be dealt twice.) With live casino roulette, it is possible to get a reasonable calculation on where the ball will land to overcome bias, while the ball and wheel are in motion and bets are still being allowed. However, It does not appear that your program is capable of detecting such things and making the needed calculation.
Also, I do not think the casual gambler is going to have enough sessions to detect a bias near the middle of a particular session. This appears especially true when it appears that your program requires the user to manually input the results. In order to detect a bias, you would need a large sample size, probably in the billions, and that is being overly generous.