I was exactly in the same place 5 years ago ... I had wasted 3 years discretionary trading several "trading plans" w/o success.
I decided to backtest my trading ideas using the computer. I took me several years to get to my first mildly profitable system, BUT I had enough faith in it that I could let the computer trade without too many interferences (I did track those, and their cost, it helped me become a really disciplined trader).
You have to go back to the drawing board, study quantitatively price-action, get good stats, figure-out from those stats one setup with a positive expectancy, and build a plan around it.
I believe it is a waste of your time to look at this chart, or that chart, or a dozen of them, even a hundred of them, and just try to see a recurring pattern. You need to create a model of price-action, process 5 to 10 years of market data into that model, then research *in the model* patterns with a statistically profitable outcome.
This isn't as much work as it sounds (at least, if you are any good at programming). The key is to have a mechanic way (software) to process the historical data into your model, so that it is 100% consistent and you can use that part later in an automated system.