blackswan-mvp

blackswan-mvp

Automated stock trading solution.

Table of Contents

1. About The Project
   • Built With
2. Getting Started
   • Prerequisites
   • Installation
   • Config
   • Usage
3. Development
4. Roadmap
5. Contact

About The Project

(back to top)

Built With

• Python
• Numpy
• Pandas
• Scikit-Learn
• Alpaca
• Simfin
• Finviz

(back to top)

Getting Started

To get a local copy up and running follow these simple steps.

Prerequisites

These commands are written assuming you are on a Linux/MacOS system.

Make sure these dependencies are setup before installing.

• Install docker with these instructions
• Set up an account with Alpaca Markets and get your API keys
• Set up an account with FinnHub and get your API key

Installation

1. Create a .env file and copy your keys like so

    # alpaca paper values
    APCA_API_KEY_ID=xxxxxxxxxxxx
    APCA_API_SECRET_KEY=xxxxxxxxxxxx
    APCA_API_BASE_URL=https://paper-api.alpaca.markets
    APCA_API_DATA_URL=https://data.alpaca.markets
   
    # finnhub key
    FINNHUB_KEY_ID=xxxxxxxxxxxx
   
    # misc
    NUMEXPR_MAX_THREADS=12
    TF_CPP_MIN_LOG_LEVEL=3
   

2. Pull the docker image

    docker pull ckinateder/blackswan-mvp:latest
   

(back to top)

Config

Edit the config file to your choosing, or create a new one. There’s a built-in linter that will help you create your config. There are two options for symbols: either auto <n>, indicating you would like blackswan to choose n symbols for you, or a list of symbols separated by spaces.

# trading
symbols: auto 4 # or list of symbols
model_expiry: 8h  # how often the model needs to be retrained
portfolio_max_value: 5000  # max holding at one time
open_order_timer_limit: 9  # how long until cancel pending order
close_on_exit: yes # close out positions on exit
rebalance: 0.01 # does nothing yet
min_eval_profit: 1.03  # backtesting must exceed x to start trading
back_padding: 7d # how far back to fill data

# logging
loglevel: INFO # one of {DEBUG, INFO, WARNING, ERROR, FATAL}
logpath: runs
modelpath: models

(back to top)

Usage

Run the container with

 docker run --env-file .env -v ${PWD}/models:/blackswan-mvp/models -v ${PWD}/runs:/blackswan-mvp/runs -v ${PWD}/configs:/blackswan-mvp/configs --rm ckinateder/blackswan-mvp:latest

The trained models will be saved in the models directory and logs in the logs directory.

usage: main.py [-h] [-c CONFIG] [-f]

CLI args for blackswan-mvp

optional arguments:
  -h, --help            show this help message and exit
  -c CONFIG, --config CONFIG
                        path to config file (default: configs/default.yaml)
  -f, --force-retrain   force retrain models (default: False)

Best Practices

I’ve noticed that the best performing symbols tend to have high volume and low volatility.

(back to top)

Development

Make sure you have Python 3.9+ installed.

1. Install TA-lib

    wget http://prdownloads.sourceforge.net/ta-lib/ta-lib-0.4.0-src.tar.gz && \
      tar -xf ta-lib-0.4.0-src.tar.gz && \
      cd ta-lib && \
      ./configure && \
      make && make install
   

2. Clone the repo

   git clone https://github.com/ckinateder/blackswan-mvp.git && cd blackswan-mvp
   

3. Install python packages

   pip install -r requirements.txt
   

4. Run the program with

   python src/main.py
   

5. Package new changes in docker image

   docker build . -t blackswan-mvp:new-changes
   

(back to top)

Websocket Messages

The trader has a websocket stream.

(back to top)

Roadmap

• Automatic symbol selection
• Capital allocation
  • Per-asset ROI
  • Rebalancing
• Advanced loss analysis
• Better config options (yaml?)

See the open issues for a full list of proposed features (and known issues).

(back to top)

Contact

Calvin Kinateder - calvinkinateder@gmail.com

Project Link: https://github.com/ckinateder/blackswan-mvp

(back to top)