mate is an experimental mini algorithmic trading engine. Use at your own risk.

I am writing this to learn more about algorithmic trading, and quantitative finance.

Mate supports the following account types:

You will need access to an API with live market data. Currently only supporting TD Ameritrade.

Configuration is passed in through environment only currently:

To register an account and get API access, you can either see upstream documentation, or try your luck with the steps below:

• Sign up for a developer account, even if you have a brokerage account

• Create your application (if you plan to run this locally, use http://localhost or http://127.0.0.1 for the redirect_url)

• go to https://auth.tdameritrade.com/auth?response_type=code&redirect_uri=<Your url encoded callback url>&client_id=<Your Client Key>%40AMER.OAUTHAP in a browser, replacing <Your url encoded callback url> with the redirect_url you created when you created your TD Ameritrade app, and replacing <Your Client Key> with the client key from your TD Ameritrade app.

  • eg https://auth.tdameritrade.com/auth?response_type=code&redirect_uri=http%3A%2F%2Flocalhost&client_id=MYKEY123%40AMER.OAUTHAP

• This should redirect you to a log in for your TD ameritrade account (brokerage account), to perform the oauth grant to give your app access to your account

• if you used localhost, it will redirect and be broken, but the code to fetch your access and refresh token will be in the redirected url query params (eg: https://localhost/?code=Your%2Code%2Here)

• take that code, and go to https://developer.tdameritrade.com/authentication/apis/post/token-0

  • enter authorization_code for grant_type

  • enter offline for access_type

  • urldecode the code you copied from the redirect earlier

    ipython -c "import urllib; urllib.parse.unquote(\"YOURCODEHERE\")"

  • enter the urldecoded string into code

  • enter your application's given client_id into client_id (should end in @AMER.OAUTHAP)

  • enter your application's given redirect_uri

  • click send

  • the JSON populated at the bottom of the page contains an access_token and a refresh_token, be sure to save the refresh_token, and set it in your environment (export it in your shell rc eg: ~/.bashrc or ~/.zshrc)

note: if you end up with a failing grant_invalid error, you might want to make sure you're unquoting your authorization code, otherwise see the Authentication FAQ

Data is represented with tick data, or OHLC candles, then serialized into JSON.

Equities:

[
  {
    "close": 39.435,
    "datetime": 1546236000000,
    "high": 39.84,
    "low": 39.12,
    "open": 39.6325,
    "volume": 140013864
  }
]

This is derived from the response of the pricehistory endpoint of the TD Ameritrade API

Crypto:

{
  "1641237335": {
    "a": ["0.169609000", "221449", "221449.000"],
    "b": ["0.169608900", "268", "268.000"],
    "c": ["0.169608900", "67.14677906"]
  }
}

JSON contains a list of epoch timestamps as strings, containing an object representing the ask, bid, and close of each tick. This model is derived from the ticker endpoint response that comes from the Kraken API

The following references may be helpful for the underlying technologies used in this project: