CryptoCast: Multi-Horizon Bitcoin Price Forecasting System

🎯 Project Objective

The primary objective of this project is to build a professional-grade Bitcoin Forecasting Dashboard that helps users visualize future price movements over three distinct time horizons (1-Day, 3-Days, and 7-Days). By analyzing 60 days of historical momentum and technical features, the system provides data-driven insights into potential market volatility using machine learning.

---

🏗️ System Architecture & Workflow

This system is built as a three-stage pipeline to ensure data integrity and model performance:

1. Data Ingestion & Preprocessing: Cleans raw Bitcoin datasets, extracts technical features (SMA, Volatility), and stores them in a high-performance SQLite database.
2. Forecasting Engine: Trains a Random Forest Regressor to map historical patterns to future prices, providing accuracy metrics like MAE and MAPE for each prediction.
3. Live Dashboard: A Streamlit-based web application that allows users to interact with the model, test custom scenarios, and view visual price trajectories.

---

🛠️ Above and Beyond: Project Highlights

While the basic project requirements were met, we implemented several advanced features to deliver a premium user experience:

• Custom UI Engineering: We bypassed standard Streamlit limitations by injecting custom CSS to create a modern, high-contrast "Fintech" dashboard with trend indicators.
• SQLite Caching Layer: Instead of re-processing CSV files on every run, we implemented a persistent database layer that makes the application run "lightning fast."
• Scenario Simulation: Users can override the current market price to see how the model reacts to sudden "market crashes" or "bull runs."
• Pedagogical Code: Every line of code was refactored with full-word variable names and step-by-step documentation for maximum clarity.

---

💾 Offline Database Storage Layer

To ensure performance and reliability, the system uses a local SQLite database (bitcoin_records.db) to store processed data:

• save_data_to_sqlite_database: Exports cleaned data frames to the permanent database.
• load_data_from_sqlite_database: Pulls data back into memory for instant analysis in the dashboard.

---

🚀 Execution Instructions

Follow these steps to run the complete system:

Phase 1: Data Preprocessing

python 01_data_ingestion_and_preprocessing.py

Phase 2: Run the Forecasting Engine (Offline Audit)

python 02_multi_horizon_forecasting_engine.py

Phase 3: Launch the Interactive Dashboard

streamlit run app.py

Phase 4: TensorFlow Scalability Audit (Deep Learning Blueprint)

python 03_tensorflow_scalability_blueprint.py

---