📊 StoxFinlytics - Deep Search Market Agent

An AI-powered research assistant that performs deep search over financial markets, combining live data, web search, conflict detection, and financial education — all in one place.

Set Up and Run

1. Clone the Repository

    uv init [folder_name] # name your folder in which the repository shall be cloned
    cd [folder_name]
    git clone https://github.com/AnxiousCodeGeek/deep-search-market-agent.git
    cd deep-search-market-agent

2. Create a Virtual Environment

    uv venv
    source venv/bin/activate   # On Linux/Mac
    venv\Scripts\activate      # On Windows

3. Install Dependencies

   pip install -r requirements.txt

4. Configure API Keys You need the following keys:

• Google Gemini API (GEMINI_API_KEY)
• Tavily Search API (TAVILY_API_KEY)
• OpenAI API Key (optional, for tracing/debugging)

Locally:

Create a .env file:

GEMINI_API_KEY=your_key_here
TAVILY_API_KEY=your_key_here
OPENAI_API_KEY=your_key_here

On Streamlit Cloud:

Put these in .streamlit/secrets.toml:

GEMINI_API_KEY="your_key_here"
TAVILY_API_KEY="your_key_here"
OPENAI_API_KEY="your_key_here"

Run Locally

streamlit run frontend.py

The app is deployed on: https://ai-market-agent.streamlit.app/

Example Research Questions

Users can ask questions such as:

• 📈 “What’s the latest stock price and trend for NVIDIA?”

• 📰 “Give me the most recent financial news about Tesla.”

• 🏆 “Who are the top performing companies in the stock market today?”

• 🎓 “What is an ETF, and how does it differ from a mutual fund?”

• ⚖️ “Are there any conflicting reports about Tesla stock performance?”

What Each Agent Does

Planning Agent → Decides which agents/tools to call for a user query.

Financial Orchestrator → The “main brain” that routes user queries to the right agents and merges results.

News Agent → Fetches the latest stock/financial news (via Tavily).

Market Agent → Fetches stock/market data from Yahoo Finance (yfinance).

QA Assessment Agent → Evaluates credibility of sources.

Conflict Detection Agent → Detects contradictory info between sources.

Education Agent → Explains finance/investment concepts in plain language.

How the Team Coordinates

• User interacts via a Streamlit chat interface.

• Financial Orchestrator is the central coordinator:

  • Routes user queries to specialized agents.

  • Ensures data is consistent and trustworthy.

  • Calls the QA Agent and Conflict Detection Agent before final response.

• Sub-agents (News, Market, Education) handle domain-specific tasks.

• Results are merged, checked for conflicts, and presented back clearly to the user.

This multi-agent workflow ensures:

• Accuracy (cross-verified data)

• Transparency (quality scoring & conflict detection)

• Clarity (simple explanations for complex financial concept