This thesis aims to extract and analyze forward-looking information embedded in derivative markets, particularly through option prices and implied volatility surfaces. Options contain rich information about market participants' expectations regarding future price movements, volatility, and tail risks. The research will develop methodologies to decode these expectations and assess their predictive power for underlying asset returns, realized volatility, and market stress events. The thesis will investigate asymmetries in implied volatility (volatility skew/smile), variance risk premium, and higher-order moment risk premia. The practical objective is to construct trading strategies and risk management frameworks that exploit the information advantage derived from options markets, with applications to portfolio optimization, hedging effectiveness, and crisis prediction.

The supervisor for this topic is associate professor Tomáš Plíhal. Detailed information about the supervisor, his publications and research projects can be found here. 

The aim of this thesis is to develop and evaluate derivative-based hedging strategies for portfolio risk management, spanning both traditional optimization methods and modern deep learning frameworks. Portfolio managers face the challenge of balancing downside protection with cost efficiency when implementing hedging programs using options and other derivatives. The research will investigate classical approaches including protective puts, collar strategies, put spreads, dynamic hedging rules, and tail risk hedging, developing optimization frameworks for strike selection, hedge ratios, and rebalancing frequencies. The thesis will then extend this analysis by implementing deep hedging strategies that use neural networks and reinforcement learning to learn optimal hedging policies directly from data, accounting for realistic market frictions such as transaction costs, discrete rebalancing, and path-dependent objectives. Comparative analysis will assess when traditional methods suffice versus when deep learning approaches provide meaningful improvements. Empirical validation will include comprehensive backtesting across different market regimes, evaluation of hedging effectiveness metrics, and practical guidelines for institutional investors seeking to implement systematic hedging programs.

The supervisor for this topic is associate professor Tomáš Plíhal. Detailed information about the supervisor, his publications and research projects can be found here. 

This thesis explores the application of machine learning and deep learning techniques to develop, enhance, and implement systematic trading strategies in financial markets. Traditional quantitative approaches rely on predefined factors and linear models, while modern ML/DL methods can uncover complex non-linear patterns and adapt to changing market regimes. The research will investigate various applications including: return prediction using ensemble methods and neural networks, feature engineering and alternative data integration, regime detection and strategy switching, portfolio construction using reinforcement learning, and signal combination frameworks. The thesis will address critical challenges such as overfitting prevention, robust cross-validation in time-series contexts, transaction cost modeling, and strategy capacity constraints. Special emphasis will be placed on developing interpretable models that provide economic insight beyond black-box predictions. Empirical work will include comprehensive backtesting across multiple asset classes with realistic trading assumptions, and comparison against traditional factor-based and technical analysis strategies. The ultimate goal is to establish best practices for deploying machine learning in quantitative trading while understanding its limitations and appropriate use cases.

The supervisor for this topic is associate professor Tomáš Plíhal. Detailed information about the supervisor, his publications and research projects can be found here. 

The thesis will investigate dependence in financial markets using spectral and quantile-based methodologies. It will examine how quantile spectral tools (e.g., quantile coherency) and risk-measure-driven approaches, emphasizing expected shortfall, can capture nonlinear, asymmetric, and tail linkages across assets and horizons. The thesis will consider flexible dependence models and compare their performance for inference, forecasting, and risk transmission analysis. Applications may span equities, rates, FX, and alternative assets, with attention to stress and tranquil regimes.

The supervisor for this topic is associate professor Tomáš Výrost. Detailed information about the supervisor, his publications and research projects can be found here. 

The thesis will analyze how participation in public support instruments like grants,  innovation and cohesion programs, and national/state-aid schemes affects firms’ financial and economic performance across a large European firm-level panel. The thesis may implement modern identification strategies for staggered treatments, such as DiD, event studies, generalized synthetic control, quasi-experiments around eligibility thresholds, and doubly robust/causal-ML estimators to uncover heterogeneous effects by ownership, size, and sector. The thesis will estimate efficiency and productivity using dynamic/network DEA and flexible SFA to separate frontier shifts from inefficiency. The thesis outcomes will concentrate on total factor productivity, profitability, investment, employment, survival, and spillovers.

The supervisor for this topic is associate professor Tomáš Výrost. Detailed information about the supervisor, his publications and research projects can be found here. 

For new information to be incorporated into financial asset prices, investors need to pay attention to that information. However, in an information-rich world, attention is a limited cognitive resource that influences investors’ behavior. The goal of this thesis is to write essays that will theoretically and empirically explain how limited investor attention might influence the behavior of asset prices. Within existing models, the dissertation should focus especially (but not exclusively) on the role of distraction and the formation of attention-grabbing information.

The supervisor for this topic is professor Štefan Lyócsa. Detailed information about the supervisor, his publications and research projects can be found here. 

The aim of this thesis is to design a prediction model of corporate financial health based on empirical data. The research focuses on examining the predictive power of well-known bankruptcy and credit risk models in the context of corporations operating in the Czech Republic and Slovakia. The main contribution lies in the development and empirical validation of an original bankruptcy model specifically tailored to these countries. From a theoretical perspective, the dissertation also verifies the applicability and predictive accuracy of established models, which, although widely recognized in the literature, were not originally constructed using data from Czech and Slovak firms.

The supervisor for this topic is associate professor Eduard Baumöhl. Detailed information about the supervisor, his publications and research projects can be found here.