Data and Codes for "Predicting Cryptocurrency Volatility: The Power of Model Clustering"

These codes replicate all the empirical results presented in the paper "Predicting Cryptocurrency Volatility: The Power of Model Clustering" by Yue Qiu, Shaoguang Qu, Zhentao Shi, and Tian Xie.

1. Input Data: crvg_data.mat
This file contains the primary dataset used for the empirical analyses in the paper.

2. Main Script File: crvg_main.m
- This script performs all the empirical exercises described in the paper.
- It reproduces Figures 1, 3, 4, and 5 in the main manuscript (excluding Figure 2, which is a flowchart), as well as Figure A.1 in the appendix.
- It also replicates Tables 1 to 10 in the main manuscript and Tables A.2 and A.3 in the appendix (Table A.1 lists selected representative cryptocurrencies).

3. Function Folder: \functions\
This folder contains 17 essential function files required to replicate the empirical results.

4. Result Files: result_main.mat, result_robust1.mat, result_robust2.mat, result_robust3.mat
- Forecasting exercises can be time-intensive, so saving intermediate results is recommended before conducting further analysis.
- Pre-saved result files are provided for the main analysis and three robustness checks.
- While the provided codes can generate these result files, the process may take some time.
- The forecasting and testing procedures involve inherent randomness, which may lead to slight variations in quantitative results. However, the qualitative conclusions remain consistent.