the output in which I can test new trade ideas is really important to me.

if I have to dig the entire codebase each time I want to test a new feature, market, alternative dataset, that loss of time compounds and good ideas get left behind.

this has been my focus to fix and I'm really close to a level of efficiency that I am happy about.

it has to be as close to a click of a button as possible.

obviously the feature engineering requires some adaptation and also if different market types are introduced.

but all else should not have to be re-adjusted on each iteration for the most part.