Lookahead for Strategy002 and 004

[froggleston]

Lookahead for Strategy002 and 004

[froggleston]

Happy to close this.

[xmatthias]

with Strategy004 it's clear - we'll be fixing this.

for the "hammer" issue - i'm not sure.
I've never used pattern functions before - so i'm not really certain on how they work.

The same goes with the function you point out - i struggled to find the definition for it - so it may just use "everything up to now" ... which would kinda be fine (it's not lookahead per se) - though it'd still be a form of problem, as it assumes an (indefinitely) expanding window.

[froggleston]

I'm happy to raise a PR for 004.

Yeah, with the pattern functions, it's unclear as to what an empty lookback function actually does in talib...

[xmatthias]

well technically - i think the lookback is pretty much given by the indicator itself.

a hammer is just 1 candle - in a pattern, something between 3 and 5 (depending if you wait for confirmation) - so even if it were to use expanding windows - a candle 200 lines from the pattern wouldn't count.

I'll leave it up to you if we want to pursue the hammer problem further, or if we assume #285 to close this issue (which means it could be closed).

[froggleston]

Yeah I started a thread in the main discord about it for the candle pattern stuff, but mainly focused around these two SO posts:

• https://stackoverflow.com/questions/61510964/how-to-set-custom-lookback-in-ta-lib-pattern-recognition-for-python
• https://stackoverflow.com/questions/50358479/ta-lib-technical-analysis-library-lookback-and-unstableperiod

It seems like the lookback isn't problematic, just completely impossible to find without looking into the code. It's not documented anywhere easy.