Ledger Instances

Overview

Ledger instances allow the ML team to forward-test their algorithms. An API and database are hosted on the PC that allows members to create instances of Ledgers and tie specific algorithms to trade on them.

Architecture Diagrams

User Endpoints

Model Interaction

API Features

1. create_ledger To create a ledger.

   Expected arguments:

   • name: unique name of algorithm
   • tickerstotrack: two tickers (e.g. (AAPL, GOOG))
   • algo_path: path to algorithm from the projects directory (ex. algo_path=harv-extension)
   • updatetime: time interval for updates (minutes)
   • end: lifespan of instance (days)

   Example command: https://watstreet/create_ledger?name=krishalgo&tickerstotrack=AAPL,GOOG&algo_path=https://github.com/Wat-Street/money-making/tree/main/projects/ledger_test_model&updatetime=1&end=100

2. view_ledger To retrieve details of a specific ledger.

   Expected arguments:

   • name: name of ledger.

   Example command: https://watstreet/view_ledger?name=krishalgo

3. delete_ledger To delete a ledger.

   Expected arguments:

   • name: name of ledger.

   Example command: https://watstreet/delete_ledger?name=krishalgo

4. update_ledger (Protected Endpoint) To update a ledger's trades and holdings. Requires API key in X-API-Key header.

   Expected arguments (JSON body):

   • name: name of ledger
   • trades: list of new trades
   • holding: updated holdings dictionary

   Example command:

   curl -X PATCH https://watstreet/update_ledger \
        -H "Content-Type: application/json" \
        -H "X-API-Key: your-api-key" \
        -d '{
          "name": "krishalgo",
          "trades": [{"type": "buy", "ticker": "AAPL", "price": 176, "quantity": 8}],
          "holding": {"AAPL": 8, "GOOG": 0}
        }'

5. start_ledger (Protected Endpoint) To start a ledger instance. Requires API key in X-API-Key header.

   Expected arguments:

   • name: name of ledger to start

   Example command:

   curl -X GET https://watstreet/start_ledger?name=krishalgo \
        -H "X-API-Key: your-api-key"

Interaction with Models

The two standard commands are:

• trade(): runs the algorithm and, based on current ownership of stocks and balance, will return a trade
• update_book_status(book_status): gives the trading algorithm the info it needs to make trades (portfolio and balance)

Thus, models' interfaces should look like so:

class Model:
    def update_book_status(self, book_status):
        """Gives the trading algorithm the info it needs to make trades (portfolio and balance, retrieved from the view_book endpoint)"""
        pass

    def trade(self):
        """Runs the algorithm and, based on current ownership of stocks and balance, will return a trade. The result of this trade is updated into the Ledger."""
        pass

Some considerations:

• It is the responsibility of the algorithm to not violate the ledger (ie. sell more than you own or buy more than the money you have)
• In case the algorithm violates the ledger, return an exception. The algorithm must be able to deal with these standard exceptions.

Data Formats (TODO:)

KEEP THIS UPDATED AT ALL TIMES.

Basic Model Output

{
  "trades": [
    {"type": "buy", "ticker": "AAPL", "price": 176, "quantity": 8},
    {"type": "sell", "ticker": "AAPL", "price": 157, "quantity": 7},
    {"type": "buy", "ticker": "GOOG", "price": 112, "quantity": 9}
  ],
}

View Book

{
  "trades": [
    {"type": "buy", "ticker": "AAPL", "price": 176, "quantity": 8},
    {"type": "sell", "ticker": "AAPL", "price": 157, "quantity": 7},
    {"type": "buy", "ticker": "GOOG", "price": 112, "quantity": 9}
  ],
  "holding": {
    "GOOG": 9,
    "AAPL": 1,
  },
  "value": [
        "1999-01-08 04:05:06": 100000,
        "1999-01-08 04:10:06": 100800,
        "1999-01-08 04:15:06": 109000,
  ],
  "balance": 96,485
}

Design Components (wip)

1. API Backend (Flask)
2. Database (PostgreSQL, in Rebbi's local env) Database name: postgres for now Tables: (wip)

• order_books_v2
• trades

3. Scheduler (Celery?) (wip)