PNL Tracker

The tracker has been hosted on my personal website.

Demo URL : https://pnl-tracker.demo.ivorynoise.com/docs

Flow

To test the flow manually using swagger:

1. Please see the results using /trades/ & /portfolio endpoint. I have added same test data to it as mentioned in pdf :)

The data can be seen here: https://pnl-tracker.demo.ivorynoise.com/api/v1/trades

1. Please clear all datastore before hitting any route.

Design Choices

Choice 1: I implemented average-cost as the primary PnL method for simplicity and clarity

Logic is presented here

pnl-tracker/src/position/models.py

Line 51 in 978053f

def _add_long(self, position: Position, trade: Trade) -> None:

Choice 2: Use of Decimal class to avoid classical floating point errors that happen

I avoided any custom class/conversion as I wanted to focus on clarity and keep the excercise short.

Choice 3: Use of singleton classes

All datastores were mimicked using singleton classes

Choice 4: Wrote tests for Positions

Test coverage was considered only for positions as it is where the core logic lives

Choice 5: No real time updates

The market price can be mimicked using /api/v1/prices endpoint. Also, I did not optimise for real-time pnl updates. The design and engineering decisions shown here will change for such features.

Choice 6: Fee Avoidance

I did not consider fee while computing pnl

Choice 7: Supports both long and short

In case the sellQty > buyQty, we allow shorting to showcase the pnl calculation.

Setup Guide

1. Build the program using ./build-run.sh. This builds and pushes to registry. It can then be deployed via kubernetes.

2. To run locally, we can use the following command:

docker run -p 8000:8000 ivorynoise/pnl-tracker