Add a Docker / containerisation setup for reproducible deployment

[Codex-Crusader]

Problem

There is no Dockerfile, docker-compose.yml, or container configuration of any kind. Users must manually configure a Python environment, which is error-prone (especially given the unpinned dependencies — see issue #5). The README shows a "LINK TO LIVE DEPLOYMENT" which suggests the app is deployed somewhere, but there is no configuration file to reproduce that deployment.

Impact

• Onboarding friction: contributors need to debug environment issues before they can run anything
• Deployment is not reproducible — different machines produce different results
• No isolation between the app's dependencies and the host system
• No documented path for self-hosting

Suggested Dockerfile

FROM python:3.11-slim

WORKDIR /app

# Install dependencies first (layer caching)
COPY requirements.txt .
RUN pip install --no-cache-dir -r requirements.txt

# Copy application code
COPY . .

# Streamlit config
ENV STREAMLIT_SERVER_PORT=8501
ENV STREAMLIT_SERVER_HEADLESS=true
ENV STREAMLIT_BROWSER_GATHER_USAGE_STATS=false

EXPOSE 8501

CMD ["streamlit", "run", "dashboard.py", "--server.address=0.0.0.0"]

Suggested docker-compose.yml

version: "3.9"
services:
  dashboard:
    build: .
    ports:
      - "8501:8501"
    volumes:
      - ./market_data:/app/market_data  # persist snapshots
    restart: unless-stopped

Additional Considerations

• Add .dockerignore to exclude market_data/, .venv/, __pycache__/, .git/
• Add HEALTHCHECK to the Dockerfile for production use
• Document Docker usage in the README with a "Quick Start (Docker)" section
• Consider a multi-stage build to keep the final image smaller

Acceptance Criteria

• Dockerfile exists and builds successfully
• docker-compose.yml exists with volume mount for market_data/
• .dockerignore excludes non-essential files
• README includes Docker quick-start instructions
• Dashboard is accessible at http://localhost:8501 after docker-compose up

[Codex-Crusader]

Note: Docker will be introduced as a deployment layer for now, not a development environment.

The project is currently in an active experimentation phase with frequent changes.
To avoid slowing down iteration, local development will continue using a standard
Python environment (venv).

A Docker setup will be added to support:

• reproducible deployment
• easier onboarding for users
• simplified self-hosting

This will be addressed once the dependency set and core pipeline stabilize.

[Codex-Crusader]

Update: A minimal Docker setup has now been added and is available for deployment.

Current status:

• Dockerfile is present and builds successfully
• .dockerignore is configured
• Docker is intended strictly as a deployment layer (not for development)

Given the project is still evolving rapidly, this implementation is intentionally minimal:

• No compose-based workflow yet
• No production hardening (healthchecks, multi-stage builds, etc.)

Next steps (out of scope for this issue for now):

• docker-compose for easier local orchestration
• image optimization and healthchecks
• tighter integration with pinned dependencies and CI

Keeping this issue open to track those improvements, but the core requirement (basic Docker deployment) is now in place.