Hi Nikos, thanks a lot for your answer.

all clear on point 1 and 2. About time indexing, I have a timestamp index “date” in my dataset .csv (quartely-end of month date, like 31–12–2021, 31–03–2022 ecc), but is just an index..so you would suggest to insert a column called like quartely with values 12, 3, 6, 9 ecc and another one years with 2021, 2022 ecc, is it right? why can’t I just use only one column, the real column “date” as it is with complete timestamp?

also such new features, do they fall into the category type i) (temporal data with known inputs into the future)? I guess so, since if I say I want to predict 5 samples into the future I can easily say what will be the date 5 samples ahead.

I have seen the volatility.py file in the github directory of google but I would have needed a real example (notebook or colab) with volatility example to easily follow the implementation..do you have a reference for it?

Still massive thanks for your past and hopefully future replies.

Regards

Luigi