We can read in an excellent article "Clone Army" , how time consuming it is to find good candidates strategies among billions of possibilities.
In this article, we can see good Out Of Sample results, but it is taking a lot resources over long period of time. It can takes months to build a good portfolio.
We could build strategies faster if we build strategies programmatically selecting the most efficient blocks:
On one hand, we can find the most profitable building blocks of successful strategies programmatically and select the best building Blocks to Build good strategies.
(https://strategyquant.com/codebase/are-any-building-blocks-better-than-the-other-can-we-find-better-blocks-for-our-settings-for-our-instruments-and-timeframe-v2/)
On another hand, we can retest programmatically strategies but we can not build strategies programmatically
We need to be able to build strategies programmatically, selecting the best blocks, the period, the different options programmatically.
With this feature, the Custom Projects will find more strategies in a shorter period of time from the most profitable Blocks
Clone Army:
(https://strategyquant.com/blog/clone-army-development-and-running-of-a-profitable-algorithmic-trading-system/#_ftn9)
Retest programmatically:
https://strategyquant.com/doc/programming-for-sq/backtesting-strategy-programmatically-including-robustness-tests/
https://strategyquant.com/doc/programming-for-sq/running-optimizations-programmatically/
Finding the most profitable building blocks of successful strategies programmatically :
https://strategyquant.com/codebase/are-any-building-blocks-better-than-the-other-can-we-find-better-blocks-for-our-settings-for-our-instruments-and-timeframe-v2/
You are doing walk forwards and comparing the blocks of strategies performance oos?
Will new "best" blocks be selected for each workflow or each build task?
Will "find best blocks" be it's own task that we would run before a build task? If so then I guess it will have all the options of a build task + all the options of a wf tasK?
Or does it just search through a list of strategies in one databank that passed all our own filters and then compile a list of best blocks?
Building block statistics - Strategy Quant
We have now the solution for this question
We can consider this report as 'fixed'
Subject changed from [Build 136 RC1] Strategies mining efficiently : Building strategies programmatically to [Build 136 RC1] Fixed : Strategies mining efficiently : Building strategies programmatically