Volatility forecasting of green and non-green cryptocurrencies. Is there a difference? : A quantitative study evaluating GARCH-models forecast accuracy
Abstract: Cryptocurrencies are rapidly growing. The energy consumption required to be mined is huge but differs between mining processes. This thesis proposes a definition of green and non-green cryptocurrencies depending on the mining process. Green cryptocurrencies are mined through Proof of Stake and non-green are mined trough Proof of Work. Volatility investigation is an important tool for financial investors to understand and predict the market. The GARCH models have been used to forecast volatility of cryptocurrencies in several studies with inconsistent results. No research has been found that compares volatility forecasting between green and non-green cryptocurrencies. The aim of this thesis is to investigate if there is a difference in the best volatility forecasting model between green and non-green cryptocurrencies. This is done by fitting four types of GARCH models, on twenty green cryptocurrencies and twenty non-green cryptocurrencies. The models are GARCH, IGARCH, EGARCH and GJR-GARCH. The optimal mode selection is chosen based on BIC values. Forecasting is made through an expanding window with one day ahead. Forecasts are then compared with estimated realized volatility using a high and low proxy and evaluated with four loss functions: MSE, MAE, MAPE and QLIKE. The results imply that there is no difference in volatility forecasting of green and non-green cryptocurrencies. The superior model is shown to be the IGARCH followed by the EGARCH. Structural breaks are found in the data and there are small differences between the model’s predictability according to all loss functions. Therefore, results should be taken carefully.
AT THIS PAGE YOU CAN DOWNLOAD THE WHOLE ESSAY. (follow the link to the next page)