Seyed Fakhrodin Fakhrehosseini, Dr Meysam Kaviani,
Volume 15, Issue 55 (5-2024)
Abstract
Predicting financial asset volatility is highly important because this information can help investors make more informed decisions regarding buying and selling. Accurate predictions can also reduce financial risks and identify profitable opportunities. Ultimately, the ability to forecast market changes improves portfolio management strategies and minimizes unexpected losses for investors. This study examines and predicts Bitcoin price volatility by using innovative data analysis models. The Heterogeneous Autoregressive (HAR) model and its variants were selected as the primary tools for modeling volatility because of their high capability to analyze volatility data across different time scales. Given the unique characteristics of cryptocurrency markets and rapid, unpredictable price fluctuations, the use of models that can simultaneously capture both short- and long-term volatility is of significant importance. In this study, high-frequency historical Bitcoin price data from 2018 to 2022, covering 60-minute, daily, weekly, and monthly intervals, were analyzed using the HAR, HARJ, HARQ, and HARQJ models. The results indicate that heterogeneous models have strong predictive power for Bitcoin price volatility, and incorporating jump factors into these models further improves their forecasting accuracy.
Predicting financial asset volatility is highly important because this information can help investors make more informed decisions regarding buying and selling. Accurate predictions can also reduce financial risks and identify profitable opportunities. Ultimately, the ability to forecast market changes improves portfolio management strategies and minimizes unexpected losses for investors. This study examines and predicts Bitcoin price volatility by using innovative data analysis models. The Heterogeneous Autoregressive (HAR) model and its variants were selected as the primary tools for modeling volatility because of their high capability to analyze volatility data across different time scales. Given the unique characteristics of cryptocurrency markets and rapid, unpredictable price fluctuations, the use of models that can simultaneously capture both short- and long-term volatility is of significant importance. In this study, high-frequency historical Bitcoin price data from 2018 to 2022, covering 60-minute, daily, weekly, and monthly intervals, were analyzed using the HAR, HARJ, HARQ, and HARQJ models. The results indicate that heterogeneous models have strong predictive power for Bitcoin price volatility, and incorporating jump factors into these models further improves their forecasting accuracy.
