I think everybody here is showing proof of support and EXTREME patience.
And each and every one of us keep saying: stop changing 5 things at once, scream "it works", yet return 5 minutes later with another issue.
I get it ... It's not an easy topic. But hoping to solve a 20 variable system at every working point by randomly tweaking 5 variables every 5 minutes is NOT the right way to do it. It will work, for a short time, under some very specific conditions, sure.
But it seems you are reading like 10% of our messages. And ignoring the rest of it.
I already outlined my thinking about you not properly "topping off" the cells at the end (or beginning) of each day, thus your SOC keep drifting away. Coulomb counting is NOT accurate. Integrate a 0.1A difference/offset over a long period of time and it will get wrong.
@MrPablo also outlined a clear method to follow so that we can better help you.
@Der_Hannes and
@Sleeper85 are also trying to help you, whenever they have clear and enough information available to them.
Yet you seem to insist of attacking those that are trying to help you. That's just wrong