-
FinRL
-
Python 3.7
-
Alpaca SDK
First, a disclaimer - Do NOT invest any money in any type of trading bot or algorithmic engine that you are not willing to lose.
Open Final_Project.ipynb in google collab mount it and then continue.
The google collab notebook Pulls trading data for (DJI) with Yahoo Finance Downloader API It Creates a simulated trading environment using real trading data with FinRL Trains an neural network to predict that Stock Price using reinforcement learning inside this simulation with FinRL Once trained, backtests the predictions on the past years data to compute potential returns with FinRL
Unfortunately there are some problems and discrepancies with FinRL and the Backtesting software, so that I wasnt able to get all the insights, and the Google Collab notebook started crashing multiple times, which lost the data.
This will require a lot of work to get all of the insights on point, but the Neural Network Machine learning model works and produces a well trained modell, which need minor improvements in its parameters.
The File has to be imported into a Google Collab Notebook, where it will be mounted and then executed.
Unfortunately due to the fact that my venv did not work and both ignacio and sandra couldnt make it work in my notebook, I had to use google collab, which crashed multiple times during the calculation processes, which then made the calculations nearly impossible and lengthy. Therefore I was only able to run a couple of models and different hyperparameters.
Everytime i changed something in the code i had to rerun the whole think e.g. recalculate the models which take a couple of hours.
I hope you take this into account when grading my notebook.