Publication details
Modeling tail-dependence of crypto assets with Extreme Value Theory – Perspectives of Risk Management in Banks
| Authors | |
|---|---|
| Year of publication | 2022 |
| Type | Article in Periodical |
| Magazine / Source | Risk Governance and Control: Financial Markets and Institutions |
| MU Faculty or unit | |
| Citation | |
| Web | Open access časopisu |
| Doi | http://dx.doi.org/10.22495/rgcv12i4p5 |
| Keywords | Crypto Assets; Extreme Value Theory; Backtesting; Basel Traffic Light Approach; Historical Simulation; Variance-Covariance Approach |
| Attached files | |
| Description | Cryptocurrencies show some properties that differ from typical financial instruments. For example, dynamic volatility, larger price jumps, and other market participants and their associated characteristics can be observed (Pardalos, Kotsireas, Guo, & Knottenbelt, 2020). Especially high tail risk (Sun, Dedahanov, Shin, & Li, 2021; Corbet, Meegan, Larkin, Lucey, & Yarovaya, 2018; Borri, 2019) leads to the question of whether the methods and procedures established in risk management are suitable for measuring the resulting market risks of cryptos appropriately. Therefore, we examine the risk measurement of Bitcoin, Ethereum, and Litecoin. In addition to the classic methods of market risk measurement, historical simulation, and the variance-covariance approach, we also use the extreme value theory to measure risk. Only the extreme value theory with the peaks-over-threshold method delivers satisfactory backtesting results at a confidence level of 99.9%. In the context of our analysis, the highly volatile market phase from January 2021 was crucial. In this, extreme deflections that have never been observed before in the time series have significantly influenced backtesting. Our paper underlines that critical market phases could not be sufficiently observed from the short time series, leading to adequate backtesting results under the standard market risk measurement. At the same time, the strength of the extreme value theory comes into play here and generates a preferable risk measurement. |