binary.com bot machine learning algorithm ai

The task is to improve an algorithm to choose best bet size knowing this: If I win, I win only 9%, if I loose, I loose 100%. Wining ratio: 91% (at least 100 trades). I have some data sample, to be studied, but I already have studied this: - Over 116 trades, wining ratio is always 91%, before this, wining ratio is between 85% and 100% - Loses in 677 trades ocurred in this way: Lose number - Number of trades after last lose 1 17 2 4 3 7 4 4 5 8 6 2 7 3 8 2 9 9 10 7 11 4 12 78 13 2 14 11 15 6 16 3 17 5 18 2 19 3 20 21 21 14 22 6 23 1 24 20 25 4 26 1 27 9 28 23 29 3 30 7 31 11 32 20 33 25 34 8 35 0 36 14 37 4 38 14 39 15 40 3 41 18 42 3 43 18 44 9 45 2 46 19 47 9 48 2 49 21 50 5 51 20 52 6 53 25 54 2 55 1 56 16 57 19 58 0 59 12 60 3 61 5 Number of trade after last lose - Times I lose in same distance trade 12 1 15 1 16 1 17 1 23 1 0 2 8 2 11 2 18 2 19 2 21 2 25 2 1 3 5 3 6 3 7 3 14 3 20 3 9 4 4 5 2 7 3 7 So I was thinking I needed one or more machine learning algorithm to see how much i must bet every time. At this moment I choose a 20% of balance on 1,2,3 and 4th bet after last lose. I calculate a coeficient number equal to 60% from last lose trade and then I after 5,6,7,8 y 9, i multiply that coeficient to the bet, so it goes exponential decrease, thinking is more probable a new lose. It worked, but not as expected, so I go up 2x, and then 0.5x, and so on, so I want it always grows. When I lose, I use martingale algorithm, but some times, it stop everything. So I want to avoid martingale but use great success rate to create an advanced algorithm to win. My idea is to share my work already done with someone interested in improve it (it's not cheap algorithm - bot), then share me their improvements backs as coworkers. If you have the skills, you're welcome to work together a couple of days until work is done, and we make money with this. Let me know your toughts.