Finviz Long Trader

A two-part trading scaffold for a Finviz Elite screener strategy with a brains service (signal + state + orchestration) and an execution service (paper broker now, real broker later).

Features

• Minute-level polling of a configurable Finviz Elite screener.
• Strategy: buy $1,000 per symbol (configurable), create 4 staged limit-sell targets (+10%, +20%, +50%, +100%).
• Clean separation between brain (what to trade) and execution (how to trade).
• Pluggable Finviz client, market data provider, and broker interface.
• Paper broker simulation with deterministic synthetic quotes for local development.

Getting Started

1. Python 3.11+ and pip available.
2. Create a virtual environment:

python -m venv .venv
. .venv/Scripts/activate
pip install -r requirements.txt

3. Copy .env.example to .env and adjust values if needed (e.g., Finviz cookies, state file path).
4. Run the brains service (includes scheduler):

python -m src.brain.brain_service

5. Run tests:

pytest

Architecture

• src/brain: strategy, state, Finviz client, scheduling, orchestration.
• src/execution: broker abstraction and paper implementation.
• src/shared: logging setup, time utilities.
• tests: basic coverage for strategy, broker, and Finviz parsing.

Notes & Assumptions

• Finviz Elite may require authentication; provide cookies via environment variables if needed.
• Paper fills:
  • Market buys fill on the next simulate_minute call at last * 1.001 to approximate the bar high.
  • Limit sells fill when the minute mid-price meets/exceeds the limit.
• State persistence uses a JSON file for v1 and can be swapped for a database later.

Next Steps

• Wire a real market data feed and broker implementation.
• Harden Finviz scraping (or replace with an API).
• Add richer risk management, stops, and intraday resets.