Pairlistfilters in backtesting mode

[myNextCoder]

Describe your environment

• Operating system: ____
• Python Version: _____ (python -V)
• CCXT version: _____ (pip freeze | grep ccxt)
• Freqtrade Version: 2021.10 (freqtrade -V or docker-compose run --rm freqtrade -V for Freqtrade running in docker)

Your question

According to the online documentation more than one pairlist filter is supported in backtesting: Are the filters related to the past data, or only to the presence?

Ask the question you have not been able to find an answer in the Documentation

[xmatthias]

Filters usually use current data - so they should be used with care - and i'd recommend using a static pairlist (maybe one with random order) for repeatable backtesting results.

For some (like spreadfilter), it wouldn't even be possible to simulate this with historic data (historic spread info is not available).
others might have that possibility - but to "really" simulate proper behaviour, we'd need to recalculate the pairlist every few candles - which would be a huge resource hog (making backtesting slow) - as well as make the code a lot more complex than what it already is.

[myNextCoder]

I thank you for your information. In fact, I even got a repeated runtime error during backtesting (Windows Server 2022 and Miniconda), with active pairlistfilters. Since now I know, that they use current data, it makes no sense anymore for me, to use them for backtesting.