The aim of this study was to determine that portfolio diversification in different cryptocurrencies can significantly improve investment performance. It was examined whether Markowitz's portfolio optimization model is the best model for investing. In the analysis, the cryptocurrency market was taken as a sample. In the theoretical part, the cryptocurrency market is explained, the characteristics of cryptocurrencies are stated, and the modern portfolio theory, ie Markowitz's model of portfolio optimization, is presented. Value risk - VaR and conditional value risk - CVaR are further explained as alternative risk measures when creating a portfolio. In the empirical part of the paper, nine cryptocurrencies were taken in the sample, in the observed period from 02.01.2017. until 11.02.2020. The normality of the distribution of cryptocurrency yields in the observed period was examined as a basic assumption of the Markowitz model. After analysis, it was found that the yields were not normally distributed. According to the second hypothesis, optimal portfolios were created according to the Markowitz model, in order to be compared with the cryptocurrency model of equal shares. After the analysis, it is concluded that Markowitz's optimization model gives better results than the cryptocurrency model of equal shares, ie by investing in the portfolio according to Markowitz's model, the investor achieves higher returns. Furthermore, according to the third hypothesis, Markowitz's model was compared with the CVaR-based model, in order to determine which risk measures the investor achieves higher returns. According to the results, the CVaR-based model gives better performances than the Markowitz model.