Back tested systems don't work they suffer from data-snooping bias. look it up on wikipedia.
I need to disagree with you or if you like I just partially agree with you. If you carefully read this article
it says that the 'MISUSE of data mining techniques that can lead to bogus results'. In our case this is the backtesting.
I totally agree with it, which means that if you write an EA with a certain number of parameters (more is better),
and start to backtest it heavily, I'm sure you will find some setups that results a positive balance. In this case you
hardcode / hardwire the past, hoping that the future will be the same. And that's a failure. A misuse of backtesting.
But I believe that if you want to develop a successful EA, it MUST preform on backtesting. So it's a required criteria.
So the key is the right usage of the backtesting tool. I think my quite sophisticated development / testing framework
satisfy this criteria. But time will tell...
The strategies section is only backtested , you have not yet forward tested this system. Unless you have an account under https://www.myfxbook.com/members/bcsendes
? With back tests we can run any set of arbitrary parameters , trying out thousands of different combinations. But this set of parameters will not work in a forward test because history never repeats itself.
Your back test ended May, if you were to back test from June, July, Aug, Sept then the system will fail.
Note that I differentiate between history and repeating patterns. It is possible to time the markets but this is based on patterns within historical events. These historical events EAs can't cope with