::table-of-contents::

Pattern recognition

short-term

recognize patterns in price movements to make buy/sell decisions.

• many features: training machine learning models, using technical indicators, etc.
• simple patterns: moving averages, support/resistance levels, trend lines.

News-based trading

long-term

analyze news sentiment to predict market movements using AI.

• search for relevant news articles
• analyze sentiment (direction, intensity, probability)
• make trading decisions based on sentiment analysis

This strategy can be executed parallel to other strategies, as it focuses on long-term trends.

NEWS + AI + PATTERN RECOGNITION

• concentrate on one single stock (e.g. Tesla)
• use news-based trading for long-term trends
• use pattern recognition for short-term terms
• check if the trend falls to fast -> immediate sell
• check if the trend rises to fast -> immediate buy

Agentic Traiding

AI agent gets tools to traide and fetch news and acts autonomously.

Tools

• fetch news
• spawn agent (e.g. for sentiment analysis)
• buy stock
• sell stock
• check portfolio
• check stock price
• wait_for (time, event, condition (news, stock price, ...))
• make_note

Problem Context Length

• agent can only remember a limited amount of information
• RAG for news articles and important events
• master agent can spawn sub-agents for specific tasks (e.g. sentiment analysis)
• tool for memoization (~make notes)