This repository contains a detailed time series analysis of daily Bitcoin closing prices from September 2014 to March 2025. Using classical statistical techniques and visualizations, the project investigates trends, stationarity, volatility, and distributional properties of Bitcoin prices.
Bitcoin exhibits highly volatile behavior, making it a unique candidate for time series analysis. This project focuses on understanding the underlying dynamics of Bitcoin's daily closing prices and provides insights into:
- Price trends and seasonality
- Stationarity and differencing
- Volatility and outliers
- Residual analysis and fitted distributions
- Auto-correlation and partial auto-correlation
- Logarithmic Transformation for variance stabilization
- Moving Averages for trend visualization
- First-Order Differencing to address non-stationarity
- Augmented Dickey-Fuller Test for statistical stationarity checks
- IQR Method for outlier detection
- ACF & PACF Plots for temporal dependency analysis
- Distribution Fitting including t-distribution to model heavy tails
- Source: Yahoo Finance – BTC-USD
- Fields:
- Date
- Open, High, Low, Close prices
- Volume
- Dickey, D.A., & Fuller, W.A. (1979). Distribution of the Estimators for Autoregressive Time Series with a Unit Root.
- Yahoo Finance (2025). Bitcoin Historical Price Data.