Weird behaviour in RSI moment indicator on development branch about pandas-ta HOT 9 CLOSED

Hi @twopirllc,
thanks for such a lighting fast response. :)

Yes, you are correct in assuming that I am not using talib. I think the issue boils down to the fact the development branch is using rma as the default ma instead of sma which is on the main branch.

If I set the mamode="sma" parameter like so

import pandas_ta as pta

df[DataframeColumns.RSI] = pta.rsi(
    df[DataframeColumns.CLOSE], length=self.settings.rsi_default_length, mamode="sma"
)

the prenans are returned as expected.

from pandas-ta.

@einarjohnson

Oh yeah! It has been awhile since I touch rsi. Seems I chose to default to TradingView's rma calculation if TA Lib is not installed.

from pandas-ta.

@einarjohnson

Yes it is easy to forget changing settings over time.

Anyways, your help and input is very much appreciated and thanks for this brilliant library, it has proven most useful for our project.

Sponsorship or donations help massively to keep this project alive. 😎

Buy me a Coffee

Thanks!

from pandas-ta.

apologies for lateness here, been up to my eyeballs with life/work/kids/etc. the coffee has been delivered. ☕

from pandas-ta.

I know how it is. Much appreciated. Thank you!

from pandas-ta.

Hello @einarjohnson,

RSI works as intended with the prenans. There are no plans to alter this behavior.

Unfortunately, I am unable to replicate you results with the development branch. What is your code to reproduce the resultant output?

Kind Regards,
KJ

from pandas-ta.

@einarjohnson

Oh I see... you are not using TA Lib. Looks like you are correct with the lack of prenans for the Pandas TA RSI version.

You are correct that it should have prenans there. I'll add it to the todo list. Thanks for the heads up.

from pandas-ta.

Interesting. SMA seems to be the default one when I use TW and RMA(labelled as SMMA?) appears to be unselected.

maybe this is a user setting or something that is bound to my account, could very well be since I use TW so much and I might have changed this at some stage (i can't remember doing that though)

Anyways, your help and input is very much appreciated and thanks for this brilliant library, it has proven most useful for our project.

from pandas-ta.

🤣

from pandas-ta.