Zero confidence in its Strategy Tester
I have over a year of experience programing in the MQL4/MQL5 languages, I have worked through many quirks and gotchas and amassed a considerable local code repository - so I am qualified to make such a bold statement.
Today I am just throwing in a towel. My lack of confidence in its ability to simulate trades is now absolute ZERO and casts serious doubts onto previous results.
The picture below is of a custom symbol, left is the data Strategy Tester, right is normal chart history, both of them have a signal attached. You can clearly see the Strategy Tester chart is different and missing bars and so of course the signal activates differently. I've discovered this issue when I tested my multi symbol expert on Symbol A and then on Symbol B and found the sum net profit and orders placed to be different than when I tested the same expert on Symbol A and Symbol B at same time.
If you are considering MetaTrader as a trading platform that you can grow with, just don't - save yourself the pains I've experienced.
I do not have an alternative tool yet so feel free to suggest in the comments section.
On the bright side the 'learn to trade' journey was and is still glorious and worth it.
2018-Nov-06 Update: Still a random number generator
Below is screenshot which shows genetic optimization run and below the first select to be run as a single test, note the results completely don't match.