Crypto Price Predictor

Environment

• Ubuntu 22.04
• Make 4.3
• Python 3.10
• Docker 26.1.4
• Docker Compose 2.25.1

Setup

Take trades microservice as an example in Usage section.

uv

uv: an extremely fast Python package and project manager.

Install

curl -LsSf https://astral.sh/uv/install.sh | sh

Usage

Create a project with uv.

cd services
uv init [project-name]
# e.g. uv init trades

Then the folder structure will be created automatically.

services
├── trades
│   ├── .python-version   // the Python version used in the project
│   ├── hello.py          // a simple hello world program (the entry point of the project)
│   ├── pyproject.toml    // the configuration file for uv
│   └── README.md         // the README file

Usually change hello.py to run.py and write the main logic in run.py.

Make

Make: a build automation tool.

Install

sudo apt install make

Usage

Write a Makefile in the project.

cd services/trades
make [target]
# e.g. make run

Redpanda

Redpanda: a fast, reliable, and cost-effective Kafka®-compatible event streaming platform.

Usage

Use the docker compose file in docker-compose/redpanda.yml. And run the following command.

cd docker-compose
docker compose -f redpanda.yml up -d

Pre-commit

Pre-commit: a framework for managing and maintaining multi-language pre-commit hooks.

Install

uv tool install ruff  # use ruff as the linter and formatter in pre-commit
uv tool install pre-commit

Usage

Install the git hooks scripts.

pre-commit install
# Ouput:
# pre-commit installed at .git/hooks/pre-commit

Then every time you commit, the hooks will be run automatically.

Or you can run against all the files manually, run the following command.

pre-commit run --all-files

Steps

Feature pipeline

• Ingest trades from external API
• Transform trades into technical indicators
• Save these technical indicators to a feature store