Bitcoin Price Prediction Flask App

This project is a Flask-wrapped Bitcoin price direction predictor. It collects live market, funding, macro news, AI/semiconductor stock, and commodity signals, combines them with configurable rule-based weights, and displays a short-horizon BTC forecast in a web dashboard.

The submitted version contains only the working web application and runtime files. Training scripts, model fine-tuning reports, generated model artifacts, prediction logs, caches, and monitoring logs are intentionally excluded from the GitHub repository.

Features

• Flask web interface for running the predictor from a browser.
• Selectable 1-hour and 4-hour prediction horizons.
• Live BTC market data from CoinGecko and Fear & Greed Index data.
• Binance funding/open-interest collectors with graceful error handling if unavailable.
• Optional NewsAPI-powered U.S. policy and geopolitics sentiment signals.
• AI chip stock and oil/gold factor inputs through yfinance.
• Factor breakdown, confidence label, composite score, BTC spot price, projected range, warnings, and recent local prediction summary.

Project Structure

.
├── app.py                  # Flask web application
├── script.py               # Prediction engine, collectors, scoring, and CLI helpers
├── ml_feature_contract.py  # Shared feature schema used by optional shadow logic
├── config.json             # Runtime configuration and factor weights
├── requirements.txt        # Python dependencies
├── .env.example            # Optional environment variable template
├── templates/
│   └── index.html          # Flask page template
└── static/
    └── app.css             # Dashboard styling

Preview

The locally running Flask dashboard after a real engine run using public market endpoints, followed by its factor inputs. Unavailable news factors remain explicitly marked; no API keys or trading account were used.

Requirements

• Python 3.10 or newer
• Internet access for live market/news data
• Optional: a NewsAPI key for policy and geopolitics news sentiment

The app still runs without NEWS_API_KEY; the affected news factors are marked unavailable and the remaining factors are used.

Setup

1. Clone the repository and enter the project folder:

git clone <your-repository-url>
cd <your-repository-folder>

2. Create and activate a virtual environment:

python3 -m venv .venv
source .venv/bin/activate

On Windows PowerShell:

python -m venv .venv
.\.venv\Scripts\Activate.ps1

3. Install dependencies:

pip install -r requirements.txt

4. Optional: configure NewsAPI access:

cp .env.example .env

Then edit .env and set:

NEWS_API_KEY=your_newsapi_key_here

Running the Flask App

Start the web server:

python app.py

Open the local app in a browser:

http://127.0.0.1:5050

Choose either the 1h or 4h horizon and click Run Current Engine to generate a prediction.

Optional CLI Usage

The same prediction engine can also be run directly:

python script.py --hours 4

For a 1-hour prediction:

python script.py --hours 1

The CLI may append entries to predictions.log; this file is ignored by Git and is not required for the Flask app submission.

Configuration

config.json controls:

• Default prediction horizon
• Factor weights
• Binance futures symbol
• Stock and commodity tickers
• News keywords
• Accuracy and cache-related settings

The Flask app disables optional shadow model inference at runtime, so no .pkl model file is required to run the web application.

Files Excluded From GitHub

The .gitignore excludes local/generated files such as:

• .env and API keys
• Virtual environments
• predictions.log
• news_cache.json
• open_interest_cache.json
• runlogs/
• model artifacts in models/
• tuning/report documents in docs/
• monitoring/automation files
• training and evaluation scripts

This keeps the repository focused on the runnable Flask application only excluding the fine tuning model stage.

Notes

This project is for educational use. The prediction output is a rule-based analysis of public market and news signals, not financial advice.