Additional Data for training

I added Sentiment and Macroeconomic indicators to the dataset to give the agent a better context during training. I also added data points such as volume and data repetitive loop, converted into which works better with machine learning algorithms. These fields are leveraged primarily by the code at https://github.com/mikazlopes/training-farm.

You will need to get your own API keys from AlphaVantage and EOD Historical to be able to pull the data.

0.4

Improved code from the 0.3.6 version of FINRL. Significantly improved data download and treatment speed when used in conjunction with https://github.com/mikazlopes/training-farm. It allows the cache of Alpaca downloads for faster processing. Removed several Dataframe copy operations to reduce RAM usage and speed up data cleaning. Fixed a few issues that prevented executing the paper trading demo notebook.

What's Changed

• Single threading by @mikazlopes in #1

New Contributors

• @mikazlopes made their first contribution in #1

Full Changelog: https://github.com/mikazlopes/FinRLOptimized/commits/v0.4