This project implements an autonomous AI agent that scans real-time market data via APIs, identifies high-momentum stocks from gainers lists, parses financial news using LLM (LLaMA), and classifies opportunities using custom prompts. The agent outputs actionable investment signals and risk flags with full automation, serving as a front-end screener for forecasting pipelines. Integrated with Pushover app for on-the-fly mobile phone notifications.
The results and statistics provided below were obtained with no human intervation. The only input to the code is the stock ticker list.
- Real-time market analysis. Ranking system on Trend Score, derived from LLM sentiments on news, and a Momentum Score, based on up-to-date financial data.
- Identify bull cycles based on up-to-date financial data and provides investment recommendations and rebalance opportunities.
- Fully automated in --live mode. Runs until manually stopped.
- Supports real-time mobile phone notifications via Pushover API.
- Easialy integrated with any Ollama version.
- ✅ Real-time market analysis.
- ✅ Full automation.
- ✅ Mobile notifications via Pushover.
- ✅ Easily extendable for other financial APIs.
git clone https://gitlab.com/bcucco/autonomous-ai-agent-market-opportunity-scout.git
cd autonomous-ai-agent-market-opportunity-scout
pip install -r requirements.txt
NUM_ARTICLES = Number of news to parse for each ticker.
- PHASE 1 – Ticker Screening: Identify potentially explosive short-term stock opportunities.
Pulls top gainers from market data using get_top_gainers(). Queries news headlines or summaries using your LLM-based function using ask_ollama(). Parses this response and classifies the opportunities. Saves info in a .json file.
- PHASE 2 – Deep Analysis (LLM + Manual Context): Generate a more detailed summary and reasoning behind selected opportunities.
For tickers flagged in Phase 1, it runs analyze() or summary() prompts through the LLM again. This time, the model gives deeper insight, such as a breakdown of technical indicators, sentiment or catalyst analysis from financial news and warnings or caveats to watch out for. Responses are saved separately and shown in a user-friendly way for final review.
On-screen
: All the info described in Phase 2 above.scout_data/
: Folder containing all market data for each ticker.
.
├── scout.py # Main code
├── README.md # This file
├── requirements.txt # Contains dependencies to be installed
├── scout_data/ # (Generated) Cached historical stock data
├── LICENSE # Code's License Information
This project is protected under the terms impose on its license.
See the LICENSE file for more details.
Code developed by B. Cucco.