OON Trading Bot

An automated trading bot for the OON Boersespiel (OON Stock Market Game) using AI-powered decision making, technical analysis, and browser automation.

Important

This is a research/educational project for a stock market simulation game. Use at your own risk.

Features

• Multi-Phase Trading Loop - 3-phase strategy: Defense (portfolio safety), Tech Scan (market data), Synthesis (AI + algo buy decisions)
• Multiple AI Providers - Google AI Studio (browser), OpenAI, Claude, Google Gemini API, Ollama (local)
• Technical Analysis - RSI, EMA, VWAP, ATR indicators via Yahoo Finance with momentum scoring
• Smart Risk Management - Stop-loss, take-profit, high water mark trailing stop, EOD protection
• Live Dashboard - Streamlit web UI with portfolio view, transaction history, and live console
• Web Config - Browser-based AI provider configuration page
• Auto-Recovery - Supervisor pattern with automatic restart on errors
• Transaction Logging - Daily log files with full trade details and AI reasoning

Project Structure

oon-TradingBot/
├── src/
│   ├── main.py                  # Entry point & supervisor loop
│   ├── core/                    # Infrastructure
│   │   ├── config.py            # YAML config loader & constants
│   │   ├── utils.py             # Helpers (parsing, logging, blacklist)
│   │   └── remote_manager.py    # Bot state & control (JSON-based)
│   ├── trading/                 # Trading logic
│   │   ├── bot.py               # Main bot cycle orchestration
│   │   ├── actions.py           # Buy/sell order execution
│   │   ├── ai_service.py        # AI prompts (defense & deep-dive)
│   │   └── algo_service.py      # Portfolio safety & synthesis algorithm
│   ├── services/                # External services
│   │   ├── browser_utils.py     # Playwright browser context
│   │   ├── market_data.py       # Yahoo Finance data & technical indicators
│   │   └── oon_service.py       # OON website login & depot scanning
│   ├── ai_providers/            # AI provider implementations
│   │   ├── base_provider.py     # Abstract base class
│   │   ├── google_studio_provider.py
│   │   ├── google_api_provider.py
│   │   ├── openai_provider.py
│   │   ├── claude_provider.py
│   │   └── ollama_provider.py
│   └── ui/                      # Web interfaces
│       ├── dashboard.py         # Streamlit live dashboard
│       └── config_page.py       # Streamlit AI config page
├── config/
│   └── config.yml               # All trading parameters
├── requirements.txt
└── .env                         # Credentials (not committed)

Prerequisites

• Python 3.8+
• Google Chrome installed
• Valid OON Boersespiel account
• An AI provider (one of):
  • Google AI Studio account (free, browser-based)
  • OpenAI API key
  • Anthropic (Claude) API key
  • Google Gemini API key
  • Ollama running locally

Installation

git clone https://github.com/Gretoffel/oon-TradingBot.git
cd oon-TradingBot
pip install -r requirements.txt
python -m playwright install

Create a .env file in the project root:

BOERSEN_EMAIL=your_email@example.com
BOERSEN_PASSWORD=your_password

Note

Never commit the .env file. It's already in .gitignore.

Configuration

All trading parameters are in config/config.yml:

Section	Examples
Cycle Timing	AI cycle interval (15min), quick check interval (60s)
Risk Management	Min trade volume, max invest per stock, portfolio diversity
Strategy	AI/tech weight, min final score, max candidates
Profit/Loss	Take profit (20%), stop loss (-2.5%)
High Water Mark	Trigger at 4% profit, sell on 1.5% drawdown
Technical Indicators	RSI period, EMA spans, volume ratio
Browser	Headless mode toggle

Usage

Standard Mode

cd src
python main.py

If web_config is enabled in config/ai_config.json, a configuration page opens first to select the AI provider. Otherwise the bot starts directly.

Do not close the browser windows manually - the bot needs them for web automation.

Bot Cycle

The bot runs two alternating cycles:

1. Quick Check (every 60s) - Fetches market data, checks stop-loss/trailing stop, executes emergency sells
2. Full Strategy (every 15min) - Runs AI defense check, technical scan, AI deep-dive analysis, and executes buy orders

Web Dashboard

A Streamlit dashboard launches automatically at http://localhost:8501:

• Portfolio overview with performance and peak tracking
• Open orders
• Transaction history
• Live console output
• Pause/Resume controls

Tip

Use ngrok or similar to access the dashboard remotely.

Test Mode

Set test_mode.enabled: true in config/config.yml to run without executing real trades.

Stopping the Bot

Press Ctrl+C to gracefully shut down, or use the Pause button in the dashboard.

Logs

Transaction logs are stored in logs/ with daily rotation:

logs/
  log_2026-02-10.txt
  log_2026-02-11.txt

Each entry contains: timestamp, action (BUY/SELL), stock name, ISIN, quantity, price, profit, and AI reasoning.

Session output is also logged to logs/session_live.log.

Contributing

See CONTRIBUTING.md for development setup, coding standards, and pull request guidelines.

License

This project is licensed under the GNU General Public License v2.0.