This system is experimental, and stock prices are highly volatile. Please invest cautiously. We are not financial advisors, and nothing contained in the project is financial advice.
Full writeup: https://docs.google.com/document/d/12ebXzwmPQIeAmrdfSDvvqG13CFhqh2kIGzMFy5zs4Fk/edit?usp=sharing
This project is designed to make stock recommendations based off of indicators and a ML neural net trained on previous data for each stock.
It runs by way of daily API calls first to the Alpaca API (Premium), and then crunches the data for each stock to find the most likely stocks that are under and overvalued (referedd to fallers and risers in ths program).
Then, it takes those stocks and does a full history intraday data fetch and runs neural nets on the most extreme outliers to predict the price for the following day, outputting that information to a file with the indicators and previous and predicted price.
If you want to run the demo of this program with 50 stocks, you will have to either use the provided Alpha Advantage api key in an .env file, or supply your own paid API key. The demo key has a limit of 500 calls a day, and is far to small to handle the 3500+ stocks on the nasdaq.
Fetching stocks is also a slow process, since the API is limited to ~75 calls a minute, so api calls are limited to 1 a second.
Install requirements with pip or pycharm.
If using pip, run pip install -r requirements.txt or ./Makefile init. This will install all packages listed in requirements.txt
If You update or add a new package/library, run pip freeze > requirements.txt to save the requirements for
others!
Using something like pycharm, pycharm might want to install that for you, so you can do that there.
Project entry point is running python Advisor .
You will need both Alpha Advantage and Alpaca API keys to run this program. You will need paid versions to be able to use both.
Use NumPy and Pandas to quickly solve bulk data sets and calculate the indicators.
A neural net is used for next day adjusted closing prediction.
The indicators that are used are RSI, Stoch RSI, EMA and SMA of varying time frames (7d, 14d, and 21d, 10d, 20d, 50d, 200d).
example.env has all the codes and formats to create your own .env file, and you will need to populate the API keys yourself. If you add the email functionality in, you will need to add those permissions to an account. It is suggested you make a new gmail to do that, as allowing insecure settings might leave your account at risk.
Nolan Braman
Zuocheng Wang