Assessing the accuracy of exponentially weighted moving average models for value-at-risk and expected shortfall of crypto portfolios

Market betas of bitcoin relative to a broad crypto market index vary considerably, depending on the data source and the index selected. Even greater differences are found for ether and other cryptocurrencies. An in-depth exploration of the cause of these discrepancies reveals a long-standing incorrect time-stamping of some ranking-site data, and hence also the CRIX market index. Furthermore, individual coin data from some exchanges requires adjusting for unstable prices in the ‘stablecoin’ tether. Even then, Bitfinex coin prices have de-coupled from prices on other exchanges. Is yet another Bitfinex-tether issue arising? Finally, regarding the risk analysis of coin returns, we argue that this requires highly sophisticated models. But calibrating even the simplest GARCH model is extremely difficult because they are surprisingly sensitive to the data source.