AI Stock Analysis Dashboard

An interactive Streamlit application that combines market-data visualization, technical indicators, and local vision-language-model analysis of stock charts.

The application retrieves daily OHLCV data from Alpha Vantage, builds an interactive Plotly candlestick chart, overlays selected indicators, renders the chart as an image, and sends it to Llama 3.2 Vision through a local Ollama instance.

Features

• Daily stock data loaded from the Alpha Vantage API.
• Interactive candlestick charts built with Plotly.
• Optional 20-day SMA, 20-day EMA, and Bollinger Bands.
• Local chart analysis with llama3.2-vision through Ollama.
• A Streamlit interface for selecting a ticker, date range, and indicators.
• Local inference: chart images are processed by the locally hosted model.

Architecture

flowchart LR
    A[Alpha Vantage] --> B[OHLCV preprocessing]
    B --> C[Plotly chart and indicators]
    C --> D[Chart image]
    D --> E[Ollama and Llama 3.2 Vision]
    E --> F[Streamlit analysis panel]

Loading

Technology stack

Python, Streamlit, Plotly, Pandas, Alpha Vantage, Ollama, Llama 3.2 Vision, and Kaleido.

Running locally

1. Clone the repository and install dependencies

git clone https://github.com/Gregory-mcbit/AI-Stock-Dashboard.git
cd AI-Stock-Dashboard

python -m venv .venv
source .venv/bin/activate  # Windows: .venv\Scripts\activate
pip install -r requirements.txt

2. Configure Alpha Vantage

Create a free API key at Alpha Vantage, then create a .env file:

API_TOKEN=your_alpha_vantage_api_key

3. Install and start the local vision model

Install Ollama, download the model, and start the Ollama service:

ollama pull llama3.2-vision
ollama serve

4. Launch the dashboard

streamlit run main.py

Enter a ticker, select a date range, choose one or more indicators, and fetch the data. After the chart is rendered, select Run AI Analysis to send the chart image to the local vision model.

How the analysis works

The model receives only the rendered candlestick chart and the indicators displayed on it. It is prompted to produce a buy, hold, or sell assessment followed by an explanation. The output is exploratory and is not validated against future market performance.

Limitations

• The application uses daily Alpha Vantage data rather than a real-time market feed.
• The vision model analyzes the chart image; it does not receive fundamentals, news, order-book data, or portfolio context.
• LLM output can be inaccurate or inconsistent and should not be treated as a trading signal.
• Alpha Vantage rate limits depend on the selected API plan.

Disclaimer

This project is for educational and demonstration purposes only. It does not provide financial advice.