Dynamic GAS Based Modeling for Predicting and Assessing the Value at Risk of Bitcoin and Gold

The first step in risk management in the field of investment is to calculate the variable that explains the risk accurately. One of the most widely used criteria for calculating risk is the value at risk, which has been the focus of financial researchers for the past three decades. Therefore, the aim of the present study is dynamic modeling and variable time using a technique called Generalized Autoregressive Score (GAS) to estimate value at risk in bitcoin and gold by using daily data since 2010 to 2020 and assuming the distribution of t-student. its results are compared with the results of known AR and GARCH models. The results showed that for gold models such as GAS, GARCH and AR were able to estimate the value at risk at 5% error level. Among them, the GAS model had the best performance. For Bitcoin only two models, GAS and GARCH, are suitable for estimating value at risk and GARCH model is preferable. Also, the duration of risk of value at risk errors for all three models for gold and bitcoin lacks long-term memory, indicating its reliance on financial turmoil.