Bot errors and a small suggestion for improvement

Shpilevoy wrote: Today I tested the assembly "tiger returns" ))))
looks nice + clicker doesn't double + new moment in the arrangement appeared growing rates via rebet (acceleration)
!!! problems (bugs) arose = at some point the program itself lowers its balance ;-(



4, 15, 33, 28, 30, 2, 6, 0, 7, 14, 21, 8, 1, 23, 14, 23, 23, 23, 0, 18, 35, 10, 7, 28, 31, 20, 7, 27, 24, 17, 22, 25, 27, 14, 2, 14, 25, 34, 29, 12, 18, 18, 29, 35, 29, 30, 13, 22, 31, 32, 31, 17, 7, 29, 22, 17, 6, 29, 29, 18, 24

= after a big win on the second "29" the balance is damaged and even the previous 29 win is lost

I tested a lot today, what's interesting about this error with the balance falling down

//////////////////////////////

1) it is quite rare

2) it is not clear what it depends on

3) before such a situation everything is in order, all calculations and graphs are normal, then we enter NOT ANY (!) NUMBER
at least W, at least L, at least with an output to MAX, at least without an output to max
a) the entire game schedule breaks and the balance drops
b) with the next spin, whether W or L, the graph returns to the correct position BUT ALWAYS -15 CHIPS!

4) in this particular situation 32, 31, 17, 7, 29, 22, 17, 6 ...
further continuation
a) 6(W)-14(W+++)-14(W+++) everything is correct
b) 6(W)-14(W+++)-3(L) everything is correct
c) 29(W+++)-6(W)-14(W+++) after 6 breaks, after 14 restores, but -15 chips
d) 29(W+++)-24(L)-3(L) after 24 breaks, after 3 restores, but -15 chips
e) 3(L) - 29(W+++)-14(W+++) everything is correct

I also noticed that the balance becomes inaccurate after several stops, in the previous version this was not observed, I thought it seemed so, but no... especially somehow strangely it falls into minus on the second stop, when it went into plus and in an incomprehensible place you need to collect statistics again for some reason, while in the middle of the game if you turn on the clickerBOT, it navigates better on the hot than if you start the session from the beginning... In any case, I understand that it is better to take a break after doubling and reconfigure the clickerBOT again. It was also noticed that after the first miss on a large bet, resetting and reconfiguring again helps, sometimes immediately, sometimes better after a pause, disabling a couple of atoms can also help not to fall back to the beginning...

Thank you, we are looking into this bug.
Now the question sounds like this: How does the next number on the 50+ spin affect the fact that when the game is recalculated from the very beginning, the forecast on the 12th spin begins to change?



This kind of errors are always connected with the fact that the program has a global variable with a missed initialization. After entering data on the 50+ spin, it changes its value within the algorithm and not "0" enters the new calculation cycle, but already changed by the last calculation.
We are catching, so far without result. 3500 lines of program code is not much, we will analyze to the bottom with traces and find the error.

Of great scientific interest is the fact that a difference in the forecast of just 1 chip on the 12th spin can lead to such a colossal difference in the amount of the credit already on the 60th spin!



It clearly follows from this that the variable in question is a deterrent to betting. After the 29th number played, when the credit increased sharply, the limits expanded.

The proverb Quod licet Iovi, non licet bovi definitely works.

After that, the new expanded framework is mistakenly applied to the game from the very beginning and we see the brutal consequences. The development of the internal BRM and expert is a very delicate mechanism, where the slightest inaccuracy greatly affects the final result.

Thank you Lucky! And I was wondering why the clicker hits the number with the bet if you start it with a skipped spin, I tried this in fun and the clicker worked well with the adaptive! The spin was not skipped on purpose, but when setting up the clicks "spin", "cancel" the spin worked and I did not have time to record it in the clicker and decided to watch what would happen, about 20-30 spins on the big ones the clicker hit the number exactly, I also thought maybe I should come up with an atom of shift? I did this in the priority bet and the adaptive reacted first with a shift in the spin later, and then on the spin two three five earlier, in general, it was a complete circus and it was hard to believe that this was pre-programmed into the generation...