Here is an article with math stuff in it with time series. This has been a while since I sent something out. I am not sure how this will help. Maybe you can figure out it out by sending me your opinion.
Here is a summary of the article:
In this post, we covered fractional differencing, in particular the fixed window fractional differencing variant where we can achieve stationarity in our time series while minimizing information loss. On time series with millions of data points, fractional differencing can be computationally expensive, hindering rapid experimentation or real-time deployment. We show, through RAPIDS cuDF and numba, we can achieve 100x to 10000x speed-up depending with links to various implementations from easy (100x speed-up) to more complex (10000x speed-up).
Don’t forget if you want to learn more about the ELITE membership. This includes the latest source code walkthrough of my next crypto algo trading bot for Bitcoin. Go here to learn more
Thanks Bryan
NOTE I now post my TRADING ALERTS into my personal FACEBOOK ACCOUNT and TWITTER. Don't worry as I don't post stupid cat videos or what I eat!