Summary of the thesis:

The evolution of some phenomenon over time has interested scientists for a long time. For example, the price of stocks in stock exchange markets, the weather conditions and many other phenomena in our lives are changing over time. The set of measurements of some properties of the mentioned matters in different time intervals is called time series. Incidentally, Time Series Analysis is a field of science which analyzes the statistical properties of time series with the help of mathematical and statistical techniques. Such methods have been widely applied to stock markets, particularly to analyze and forecast stock returns and volatility. However, the major aspect of my research is the application of the Time Series Analysis on relatively young cryptocurrency markets.

**Methods.** The first part of the thesis focuses on prediction of volatility in BTC markets. To that end, an extension of the generic GARCH model is proposed. Moreover, it is shown that the proposed extension improves the predictive accuracy of the generic model and leads to a better interpretation. The findings produced by the study demonstrate high accordance with the existing literature. The second part of the thesis focuses on, yet another generic algorithm called “K-Means” algorithm. In that part, a more general algorithm is proposed which encapsulates both K-means and several other clustering algorithms. Then, we apply the proposed algorithm to daily realized volatility data of BTC to determine high and low volatility periods.