backtest: multicharts
I found that in some strategies, the conditions in the signals are completely duplicated. That is not at all necessary and wastes a lot of resources.
There is another strange thing. The logic of the signals in the generated trend strategies generally doesn't quite match common sense. For example, buy when rsi is falling and short when rsi is rising. Buy when close below supertrend, and short when close above supertrend.
I'm not quite sure if the Multicharts backtest engine is working properly.
Attachment Screenshot 2021-07-09 192618.png added
Attachment Screenshot 2021-07-09 192719.png added
Attachment Screenshot 2021-07-09 192738.png added
But I also accept the point of machine. It's just about the statistics good or not, not the common sense. It's just that there were not so obvious and so many opposite conditions in previous versions.
But completely duplicate conditions appear in the signal, I think it's still a bug. I chose not quite a few Blocks.
SQX is not a human, only machine...
and why the conditions cant be duplicated... for example if you choose only few building blocks...
i have talked here many times ago about something called simplifier (which will go through the used building blocks in the strategy and keep only those which are making some job in the strategy - for example remove redundant building blocks) of strats and this very usefull tool is not implemented yet and i dong know why, because new features (where i have doubts of their benefits) are added, but why?