| Abstract [eng] |
The master’s thesis research analyses the importance of cryptocurrencies as an alternative to traditional investment assets, in terms of the importance of falling interest rates that reduce the attractiveness of traditional investments. Cryptocurrencies are recognized in the literature for their potential to improve portfolio returns and reduce risk, but their high volatility and unpredictability pose significant challenges for investors seeking investment solutions. The focus of this study is on cryptocurrency selection and portfolio optimization to identify the capabilities and limitations of such models, and the objective is to develop an investment portfolio optimization model by applying fuzzy logic methods to cryptocurrency selection. To achieve the goal of the study, the first step was to analyze the recent literature related to the topic of the final project, including applications of fuzzy logic to cryptocurrency selection and investment portfolio optimization methods. A wide range of indicators suitable for cryptocurrency valuation were analyzed, including efficiency, risk, liquidity, data reliability and market benchmarking. Following this analysis, a structured cryptocurrency selection algorithm was developed by applying 2 fuzzy logic models: Fuzzy TOPSIS and Fuzzy VIKOR. These models assess the difficulties in decision making, such as the uncertainty and subjectivity of data and opinions. These models are applied to the selected cryptocurrencies in 3 investment portfolio optimization models: Markowitz Mean-Variance (MV), Naive (Equal-Weights) and Conditional Value-at-Risk (CvaR). For the selection of cryptocurrencies, historical data and data from European Central Bank refinancing operations are used as selection indicators, and for the evaluation of portfolios: return, risk, Sharpe, Sortino and Rachev indicators are used. The results show that fuzzy logic models are effective in reducing the number of possible cryptocurrency alternatives by highlighting the best ones according to 13 indicators selected from the literature and defined in the algorithm. The analysis of investment portfolio optimization methods shows that the Markowitz MV model provides the highest expected annual return with sufficiently balanced risk compared to other models but, due to the logic of it trying to balance the returns and risk, the portfolios are not very diversified. The results of the Naive model, which is the least costly model due to its simple logic, are highly dependent on the selection process as it produces very different results, usually anticipating loss over profit and high risk. The CVaR optimization model, which complements other models by reducing risk beyond the value at risk (VaR), is particularly suited to managing extreme market scenarios. The portfolios from this model have the best balance between expected return and risk, usually achieving slightly lower returns compared to the MV model, but also at significantly lower risk. Among the fuzzy logic selection models, Fuzzy TOPSIS produces portfolios with higher expected returns and among the top 10 alternatives includes more high capitalization categories, which results in a selection where cryptocurrencies are not only more potential, but also quite secure, regarding the risk factor. The portfolios from Fuzzy VIKOR selection model are more diversified, but with higher risk and lower return, most often resulting not in a small profit but in a loss. It is observed that Fuzzy VIKOR selection is more suitable for riskier investors, as most of the top ranked alternatives are in mid or small capitalization categories. The results underline that the performance of portfolios is strongly influenced by the valuation criteria of the alternatives chosen and the selection models applied. This study demonstrates the relevance and usefulness of fuzzy logic and traditional portfolio optimization methods for constructing investment portfolios to address the challenges of the growing and high-potential cryptocurrency market. The proposed model improves the efficiency of decision-making by managing uncertainty and subjectivity in this area. For future research, a more detailed analysis of criteria that value cryptocurrencies well and in a targeted way is recommended, as well as the application of other possible fuzzy logic models. |
