I've trained a neural network as a custom indicator using tensorflow.
I've extracted it's model into python code that doesn't require tensorflow to run, but it was too big to paste into the online editor.
So I've extracted network model into the json. But I couldn't find working methods to load local files.
I've tried to export it to csv and load, but without dates it didn't work (fetch_csv).
I've implemented server that collects price data from URLs, evaluates network and returns CSV, but I can't run fetch_csv when current market data appears.
I'm a bit frustrated.
Is there a way to load abstract user data into the memory on initialization?