Search

This article investigates the connections between the prices of biofuels and many traded commodities and other relevant assets in Europe, USA and Brazil. The analysis uses a comprehensive dataset covering price data for 38 traded titles over the period 2003-2020. We utilize the minimum spanning tree approach to identify price connections in a complex trading system. Our analysis of mutual price connections discovers the major defining features of world leading biofuels markets during the period since the ground-breaking policy initiatives of the 2003 EU Transport Fuel Directive and Energy Taxation Directive. We provide characteristics of main bioethanol and biodiesel markets with respect to government policies and technical and local features of the production and consumption of particular biofuels. Despite a relatively long and dynamically evolving history of biofuels, we find that the biofuels systems in USA, Brazil and Europe do not converge towards the same pattern of relations among fossil fuels, biofuels, agricultural commodities and financial assets.

The internet has become the primary source of information for most of the population in modern economies, and as such, it provides an enormous amount of readily available data. Among these are the data on the internet search queries, which have been shown to improve forecasting models for various economic and financial series. In the aftermath of the global financial crisis, modeling and forecasting mortgage demand and subsequent approvals have become a central issue in the banking sector as well as for governments and regulators. Here, we provide new insights into the dynamics of the UK mortgage market, specifically the demand for mortgages measured by new mortgage approvals, and whether or how models of this market can be improved by incorporating the online searches of potential mortgage applicants. Because online searches are expected to be one of the last steps before a customer's actual application for a large share of the population, intuitive utility is an appealing approach. We compare two baseline models - an autoregressive model and a structural model with relevant macroeconomic variables - with their extensions utilizing online searches on Google. We find that the extended models better explain the number of new mortgage approvals and markedly improve their nowcasting and forecasting performance.

We assess the fuel-food price linkage models of the structural and of the time series nature with the main attention devoted to the time series literature. We document shifting focus from a straightforward vector autoregressive and error-correction analysis of fuel-food commodity price level co-movement towards inclusion of biodiesel and ethanol prices directly into analysis, towards consideration of both price levels and price volatilities, and towards growing sophistication of econometric methodologies and integration of econometrics of price co-movement with theoretical considerations. The key insight of biofuels price transmission literature is that the price transmission is time and market specific, evolving with the development of biofuels policies and technologies. The most prominent relationships are coming from oil to agricultural commodities and then towards biofuels. While here are also studies not finding any strong connection between biofuels and their feedstock, the literature finding biofuels driving the prices of agricultural commodities up is negligible.

In the Czech Republic, water prices have been steadily increasing while water consumption has been decreasing during the last decades. A great deal of literature is devoted to the price elasticity of water demand but to the best of our knowledge, no similar study has been devoted entirely to the Czech Republic. In the paper, we examine the short and long-run price elasticities of water demand in the Czech Republic between years 2000 and 2011. We find that the water demand is inelastic (lower than one) with respect to water price regardless of the time period and the income elasticity is on the edge of statistical significance. Our inferences are in line with previous findings.

The motivation of this thesis is to provide a basic framework for treating long-range cross-correlated processes while keeping the methodology and as- sumptions as general as possible. Starting from the definition of long-range cross-correlated processes as jointly stationary processes with asymptotically power-law decaying cross-correlation function, we show that such definition implies a divergent at origin cross-power spectrum and power-law scaling of covariances of partial sums of the long-range cross-correlated processes. Chap- ter 2 describes these and other basic definitions and propositions together with necessary proofs. Chapter 3 then introduces several processes which possess long-range cross-correlated series properties. Apart from cases when the mem- ory parameter of the bivariate memory is a simple average of the parameters of the separate processes, we also introduce a new kind of process, which we call the mixed-correlated ARFIMA, which allows to control for both the bi- variate and univariate memory parameters. Chapter 4 deals with tests for a presence of long-range cross-correlations. We develop three new tests, and Monte-Carlo-simulation-based statistical power and size of the tests are com- pared. The newly introduced tests strongly surpass the already existing one. In Chapter 5,...

For the biofuel markets and related commodities, we study their price transmission, which is in fact equivalent to studying price cross-elasticities. Importantly, we focus on the price dependence of the price transmission mechanism. Several methodological caveats are discussed. Specifically, we combine the memory robust feasible generalized least squares estimation with two-stage least squares to control for endogeneity bias and inconsistency. We find that both ethanol and biodiesel prices are responsive to their production factors (ethanol to corn, and biodiesel to German diesel). The strength of transmission between both significant pairs increased remarkably during the food crisis of 2007/2008.

This paper provides a general overview of the social, environmental, and economical issues related to biofuels and a review of economic modeling of biofuels. The increasing importance of biofuels is driven primarily by government policies since currently available biofuels are generally not economically viable in the absence of fiscal incentives or high oil prices. Also the environmental impacts of biofuels as an alternative to fossil fuels are quite ambiguous. The literature review of the most recent economic models dealing with biofuels and their economic impacts provides a distinction between structural and reduced form models. The discussion of structural models centers primarily on computable general equilibrium models. The review of reduced models is structured toward the time series analysis approach to the dependencies between prices of biofuels, prices of agricultural commodities used for the biofuel production and prices of the fossil fuels.

In this paper, we contribute to the literature on international stock market comovement. The novelty of our approach lies in usage of wavelet tools to high-frequency financial market data, which allows us to understand the relationship between stock market returns in a different way. Major part of economic time series analysis is done in time or frequency domain separately. Wavelet analysis can combine these two fundamental approaches, so we can work in time-frequency domain. Using wavelet power spectra and wavelet coherence, we have uncovered interesting dynamics of cross-correlations between Central European and Western European stock markets using high-frequency data. Our findings provide possibility of a new approach to financial risk modeling.

In the paper, we research on the presence of long-range dependence in returns and volatility of BUX, PX and WIG between years 1997 and 2009 with use of classical and modified rescaled range. Moving block bootstrap with pre-whitening and postblackening is used for the construction of confidence intervals for the hypothesis testing. We show that there is no significant long-range dependence in returns of all examined indices. However, significant long-range dependence is detected in volatility of all three indices. The results for returns are contradictory with several studies which claim that developing markets are persistent. However, majority of these studies either do not use the confidence intervals at all or only the ones based on standard normal distribution. Therefore, the results of such studies should be reexamined and reinterpreted.

The main focus of the thesis is the introduction of new method for interpretation of fractality aspects of financial time series together with its application. We begin with description of various techniques of estimation of Hurst exponent - rescaled range, modified rescaled range and detrended fluctuation analysis. Further on, we present original theoretical results based on simulations of three mentioned procedures which have not been presented in literature yet. The results are then used in the new method of time-dependent Hurst exponent with confidence intervals developed in this thesis. Moreover, we show important advantage of using the mentioned techniques together to clearly distinguish between independent, trending, short-term dependent and long-term dependent properties of the time series. We eventually apply the proposed procedure on 13 different world stock indices and come to interesting results. To the author's best knowledge, the thesis presents the broadest application of timedependent Hurst exponent on stock indices yet.

Mostly used estimators of Hurst exponent for detection of long-range dependence are biased by presence of short-range dependence in the underlying time series. We present confidence intervals estimates for rescaled range and modified rescaled range. We show that the difference in expected values and confidence intervals enables us to use both methods together to clearly distinguish between the two types of processes. Moreover, both methods are robust against the presence of heavy tails in the underlying process.

This thesis empirically examines the alternatives to traditional asset allocation strategies based on clustering mechanisms. Portfolio selection strategies utilizing hierarchical clustering are compared to the market benchmark and traditional methods: minimum-variance and equally weighted allocation, focusing on market stress periods. The allocation strategies are tested on daily stock prices of the S&P 100 index constituents from 2005 to 2021. The performance of Hierarchical Risk Parity (HRP) and Hierarchical Equal Risk Contribution (HERC) portfolios is evaluated across several market stress periods, including the financial crisis of 2007-2008 and the global coronavirus (COVID-19) pandemic in 2020. Empirical results do not prove the superiority of hierarchical clustering allocation strategies over traditional strategies in risk-adjusted performance. JEL Classification G01, G10, G11 Keywords portfolio selection, hierarchical clustering, HRP, HERC, market stress Title Cluster-based asset allocation strategies during market stress periods

Since the 1970s, economists have started studying the concept of inflation hedging as a way to protect investments. With the recent high inflation rates, investors might be interested if newly created assets such as cryptocurrencies can be effective against inflation. This thesis paper aims to find out whether thelargest crypto asset Bitcoin can be used as an inflation hedge. To answer this question, Fisher coefficient estimation and hedging demand for the US, Euro Area, and the Czech Republic for the period between November 2014 and October 2022 will be analyzed. In addition, the vector autoregressive model (VAR)will be used for the US market in the same time frame. The results showed overall positive Bitcoin returns but all three methods indicated no or negative correlation between inflation rates in three regions and Bitcoin returns. The thesis paper concludes that Bitcoin cannot be used as an inflation hedge as notall requirements are met. Keywords Cryptocurrency, Bitcoin, gold, inflation, inflation hedge, Fisher coefficient,VAR model Title Can Bitcoin serve as an inflation hedge asset in the US, Euro Area, andCzech markets?

Left-sided athletes are often perceived as better performing as they can leverage their minority status within the sports world. While various specific left-sided athletes, such as Lionel Messi and Rafael Nadal, perform at the very top of their disciplines, these might be simply non-representative outliers. The current thesis puts the hy- pothesis of left-sided over-performance to test via a battery of tests and regressions. My thesis thoroughly analyses the prevalence and the performance of left-handed/left-footed athletes across 5 differ- ent sports. As majority of the current studies are focusing only on a few performance metrics in the given sport, my work broad- ens the knowledge on the topic since it compares the performance of left-sided and right-sided athletes in many categories in order to cover a great portion of the in-game action. Furthermore, this thesis also expands the current understanding of the (potential) left-sided advantage in direct encounters between both teams and individu- als, achieving so by implementing predictive Bradley-Terry models that are based on past matches. The overall results are rather sur- prising: in the majority of the performance comparisons between left-handers/left-footers and right-handers/right-footers, no signifi- cant difference between the two...

The thesis deals with the issue of the Video Assistant Referee in football. It evaluates the consequences of its implementation in Czech Fortuna Liga on the sample of 678 matches held during two and half seasons. The results from the models designed to treat count data were compared with relevant literature. In the form of both simple and multiple regression with additional control variables was investigated the relationship between VAR and the set of match- changing incidents, including yellow cards, red cards and penalty kicks, and the relationship between VAR and errors of on-pitch referees. The terms presence of VAR, VAR interventions and VAR as the whole were di erentiated. Whereas a significant statistical association of VAR as the whole was not revealed for yellow and red cards, a 56% increase in the number of penalties associated with VAR as the whole significantly performed. Furthermore, the negative and highly significant 118% association of the presence of VAR was reckoned in the case of errors of on-pitch referees. Subsequently, the percentage decreased due to VAR interventions, however, not su ciently to reveal a negative and significant association in errors of on-pitch referees for VAR as the whole. The exception created errors based on factual decisions. JEL Classification Z21, F21, Z29,...

In this thesis, we study the impact of individual retail investors on the financial markets. We follow the GameStop retail trading frenzy from the beginning of 2021 and the retail investors aggregated on Reddit's r/wallstreetbets. The tools employed include natural language processing, wavelet analysis and vector error correction models. The results propose that the retail investor sentiment is highly susceptible to high volatility, extreme returns and frequent news coverage. Social media is shown to exacerbate these behavioural tendencies. We find evidence that retail investor sentiment is able to predict short-term returns for stocks specifically targeted by retail investors. The findings are, however, dependent on the investment horizon. Over long horizons, we find evidence for the reversal of the relationship. Lastly, while the effect of news and social media is similar in the long run, we show that Reddit sentiment, as opposed to news sentiment, is a significant predictor of retail targeted stocks in the near term. JEL Classification C55 C58, G12, G14, G41 Keywords Sentiment, Social media, GameStop, Reddit, Natural language processing, Wavelet analysis Title Gamified Stock Markets, Sentiment and Volatility: Evid- ence from the GameStop frenzy 1

Bitcoin is often compared to gold for its gold-like features such as a store of value, a limited supply, and a safe haven. However, due to Bitcoin's extreme price movements, investors might rather look for a safe haven against Bitcoin. In this thesis, we study such properties among traditional assets. Specifically, we analyze gold, oil, and stocks as safe havens for Bitcoin on a sample period from 2014 until March 2022. We find that gold acts as a strong safe haven suggesting gold's traditional role as a shelter during uncertainty holds also for this crypto asset. 1

This thesis aims to scrutinise price gaps in the stock market. The key objective is to analyse candlestick charts surrounding price gaps and determine whether any patterns accompany their presence. Firstly, the thesis briefly describes candlestick patterns, literature relevant to price gaps and Convolutional Neural Network (CNN) as the method of choice. Price gaps are studied in a 5-minute time frame in the data of all S&P 500 constituents in the years from 2015 to 2021. By feeding images of the candlestick chart into the CNN, the proposed model reaches an Accuracy of 74.2% in predicting whether a future price will be higher or lower than the price at the gap. This result can be translated into a statement that the CNN detects hidden patterns around the price gaps. Furthermore, the thesis finds that these patterns di er across individual stocks. The thesis also shows that including news sentiment in the analysis does not improve the ability to discover patterns. JEL Classification C45, C55, C88, G14, G15, G41 Keywords price gap, convolutional neural network, pattern detection, news sentiment Title Price gaps in the stock market

The work attempts to forecast the sign of the price change for cryptoasset time series through classification. The main purpose is to find evidence concerning market efficiency of the cryptoasset markets, potential trading strategies, and differences between the modelled assets. Supporting vector machines, random forests, and multilayer perceptron models are used. An additional model aggre- gates the results of the previous three. Bitcoin, Ether, XRP, and Binance Coin are the modelled cryptoassets. The input variables include transformed daily closing prices up to five lags, trading volumes, volatility, and moving averages. Random forest models perform the best, followed by supporting vector ma- chines, and multilayer perceptrons. Aggregation does not produce improved forecasting performance. The two older assets, Bitcoin and Ethereum, are found to be less forecastable than the newer, Binance Coin and XRP. Dif- ferences between the assets exist as exhibited through forecastability. Higher classification accuracies are not found to imply better trading performance. JEL Classification C15, C69, G13, G14, G17 Keywords cryptoassets, machine learning, forecasting, cryptocurrencies Title Predictive Power of Machine Learning in Cryp- toassets

The goal of the thesis is to identify factors that drive the price of Bitcoin and the hashrate of the Bitcoin network, which represents the total computing power dedicated to Bitcoin mining, and to explore the relationship between these two variables. In the Bitcoin system, four variables were assumed to be endogenous, thus for each of them, an equation was constructed. This was the case of the Bitcoin price, the hashrate of the Bitcoin network, the total transaction fees paid, and the search volume for the term "bitcoin". The system of four equations was then simultaneously estimated, utilizing the method of Two-stage least squares. Results revealed several statistically significant explanatory variables of the price and the hashrate, including the money supply of the United States dollar or the number of unique active addresses on the Bitcoin network. The hashrate was shown to drive the price positively, however, the estimated effect of the price on the hashrate was statistically insignificant. It was argued that it might have been caused by exogenous shocks affecting multiple variables, that could not be accounted for in the data. In addition, the factors affecting the hashrate were assessed from the environmental point of view, as the high environmental impact is one of the main points in the criticism...

This thesis study two promising methods used to define the multiscale CAPM - the wavelet-based decomposition and the fractal regression. Their estimates, obtained on monthly excess return on ten portfolios formed on beta in the US market, are compared in the period from November 2000 to October 2020 and, subsequently, in the period from November 1965 to October 2020. In the first period, the multiscale beta is not significantly different from the original single-scale beta for most of the portfolios. Contrary, both methods uncover significant multiscale behavior of the beta in the second period. Specifically, the high-beta portfolios have higher multiscale beta at longer investment horizons, mainly at wavelet scale 3 and scales 12-24 of the fractal regression. Overall, both methods deliver consistent results, and seem suitable for extending the CAPM with an investment horizon. JEL Classification Keywords G12, C20 CAPM, asset pricing, multiscale analysis, wavelets, fractal regression Title Investment horizon in the CAPM: A comparison of a wavelet-based decomposition and the fractal regression

1 Abstract Since 2010 in France, local and complementary currencies have grown signif- icantly, bringing the total number in circulation to 82. Most of the studies focus on the way local and complementary currencies work and their potential benefits. With the benefit of ten years of hindsight, it is time to supplement the studies with a quantitative analysis of the phenomenon and measure its impact on the local economy. A survey dealing with all the issues involved in developing a currency completes the quantitative analysis. Nevertheless, the thesis faces difficulties in collecting this data. Hence, the thesis identifies the lack of suitable tools for regular monitoring as the main reason for the data collection issue. Despite the low response rate, the collected responses provide evidence to confirm the intuition that local and complementary currencies are dependent on their framework. Indeed, based on a few responses and spe- cific cases, the thesis distinguishes critical framework-dependent elements that impact the sustainability of a local and complementary currency. Hence, the thesis emphasizes the critical role of the local consumption habits, elected rep- resentatives, relations with public bodies, and people's attachment to the local territory. Furthermore, the thesis underlies the need to sustain...

The our thesis "Using spin model to determine FTTx connectivity market po- tential in the Czech Republic", we firstly map the current landscape of the Czech broadband technology market. Additionally, we present an overview of Ising model's interdisciplinary applications. Afterwards, we describe the dy- namics of the Ising model and in particular we study the convergence tendencies of Ising model generated series as well as the spin positioning in the Ising model lattices based on the input parameters. Consequently, we assume the spins in the model to represent the fiber tech- nology and alternative technology and thus we link the Ising model, its parame- ters and outputs to the problem of fiber connectivity potential. Apart from the standard input parameters of the Ising model, we also introduce variability in terms of the distribution of the initial lattice and we define four archetypes to represent real market situations. Ultimately, we describe the sets of parameters for which the market appears to have the most potential of fiber deployment. JEL Classification A12, C6, C15 Keywords Ising model, econophysics, fiber technology, broadband connection Title Using spin model to determine FTTx connectiv- ity market potential in the Czech Republic

Within forecasting electricity pricing, we analyse whether adding various vari- ables improves the predictions, and if shorter time intervals between observa- tions enhance accuracy of the forecasting. Next, we focus on proper selection of lagged observations, which has not been thoroughly covered in the past litera- ture. In addition, many papers studied electricity prices in larger markets (e.g. United States, Australia, Nord Pool, etc.) on datasets limited in scope, with 2-3 years timespan. To address these gaps in literature, we obtain one daily and one hourly dataset, both spanning 6 years (January 1, 2015 - December 31, 2020), from four Central and Eastern European countries - the Czech Repub- lic, the Slovak Republic, Hungary, and Romania. These contain information on the electricity prices, and information on our observed added variables - temperature and cross-border electricity flows. For the forecasting, we use two different methods - Autoregression (AR) and Seemingly Unrelated Regression (SUR). The thorough selection of lagged observations, which we accustom to the closing time of the auction-based electricity market system, serves further studies as a guidance on how to avoid possible errors and inconsistencies in their predictions. In our analyses, both AR and SUR models show that...

This thesis examine the possible e ect of consumer behaviour, specifically gold sources distribution and e ectiveness in spending, on rank between DotA 2 players. Relevant literature from the past years was summarised and dataset containing more than 91 thousand data points from OpenDota API used to determine the relationship was constructed. Ordered Logistic Regression and K-Means Clustering algorithms were applied on four datasets of 12 variables divided based on the player's position and the outcome of the game, conclud- ing that only a fraction of the variation (McFadden's R2 of 6.2% and Cox & Snell's R2 11.7% on average) has been explained and cluster analysis resulted in a mean accuracy of 38.25%. Those results were supported by the correlation analysis indicating low correlation coe cient values of dependent variable rank. Despite such output, patterns were discovered across individual rank clusters from metrics acclaiming higher e ectiveness of more experienced players re- garding consumer behaviour with strong statistical significance. It has been concluded that the e ect is present, yet its magnitude by itself is not su cient to predict the regressand. The outcome implies possible presence of other vari- ables that have an e ect on the explained variable which are not included in the...

The aim of this thesis is to examine an empirical relationship between multifrac- tality of financial time series and its returns. We approach the multifractality of a given time series as a measure of its complexity. Multifractal financial time series exhibit repeating self-similar patterns. Multifractality could be a good predictor of stock returns or a factor which can be used in asset pricing. We expected that capturing the complexity of a given time series by a model, a positive or a negative risk premia for investing into "more multifractal assets" could be found. Daily prices of 31 stock indices and daily returns of 10-years US government bonds were downloaded. All the data were recorded between 2012 and 2021. After estimation the multifractal spectra, applying MF-DFA method, of all stock indices, we ordered all stock indices from the lowest to the most multifractal. Then, we constructed a "multifractal portfolio" holding a long position in the 7 most multifractal and holding a short position in the 7 least multifractal stock indices. Fama-MacBeth regression with market risk premia and multifractal variable as independent variables was applied. Multi- fractality in all examined financial time series was found. We also found a very low negative risk premia for holding "a multifractal...

The relationship between liquidity and return predictability may be an im- portant aspect to consider when investing in cryptoassets. We examine this relation using both cross-sectional as well as panel data. First, we calculate a set of predictability measures and aggregate the results into four variables. We then regress the predictability variables on a set of controls and two measures of liquidity, specifically the Amihud illiquidity ratio and the Corwin-Schultz spread estimate. The other independent variables include the logarithm of volume, turnover ratio and Garman-Klass volatility. Results from the cross- sectional analysis indicate that liquidity negatively impacts the degree of return predictability. Moreover, findings from a subset of panel data, including only 50 cryptoassets with the largest market capitalization, provide some evidence in favor of this relationship. Results from full panel data, however, present contradictory evidence. For these regressions, liquidity is found to be either in- significant or to possess a positive impact on the degree of return predictability. Altogether, we obtain mixed evidence about the effect of cryptoasset liquidity on return predictability. JEL Classification C53, C58, G14 Keywords Cryptoassets, Predictability, Liquidity, Panel data Title...

In the last years, the field of machine learning boomed. That led to its numerous forecasting applications on prices of Foreign exchange market. Re- searchers there mostly compare neural networks to linear model baselines. The contribution of this thesis consists of a comprehensive performance com- parison between two promising machine learning methods, Support vector regression (SVR) and Long short-term memory recurrent neural network (LSTM RNN), in the forecasting of six highly traded currency pairs on one minute univariate time series data. First, it analyses methods' performances in the forecasting of one step ahead price while varying input dimensions of these methods. Next, it examines how methods perform in longer forecasts, that enabled by using a recurrent version of SVR. In the first analysis, LSTM RNN outperforms SVR in most of the cases several times. Performance of SVR is robust to varying input while LSTM RNN's performance fluctuates across dimensions. In the second analysis, LSTM RNN beats SVR mostly by order of magnitude. With increasing forecasting horizon, SVR's performance gets worse and LSTM RNN's performance remains stable. 1

This work investigates the effect of the COVID-19 pandemic on the S&P 500 stock index and its eleven sectors. Employing the ARMA and the T-GARCH model on a time series of daily returns from 2018 until March 2021, we examine the impact on volatility, returns, and day-of-the-week effect during the stock market crash caused by the pandemic and the period after. Our main findings imply that in the case of returns, the Monday effect was more negative than the Friday effect during the market crash and vice versa in the rising market after the crash. Concluding that the calendar time hypothesis holds for the observed periods. In terms of volatility, it drastically increased across the US stock market during and even after the crash. The increase was especially noticeable for the IT and Energy sectors. We also found the U-shaped daily volume pattern changed significantly with proportionately less volume of trades happening in the first half-hour of trading and more throughout the whole day.

The aim of this thesis is to investigate the Bitcoin mining profitability throughout the years 2014 to 2020 with the focus on the year 2020. The analysis is based on the break-even electricity price estimates which are obtained by using a set of variables entering the Bitcoin mining process such as the block reward, transaction fees, network hash rate or power consumption. The calculations are performed under the assumption that miner owns the most efficient mining hardware available at the time while disregarding the original investments in the necessary hardware. To further examine the relationship between the estimated break-even electricity price and the Bitcoin market price the cointegration analysis is performed employing a vector error correction model as the series seem to be nonstationary. The final results illustrate the substantial effect the Bitcoin market price has on the break-even electricity price estimates to the extent that there is rather long-term reaction in the break-even electricity price values to the shocks in the Bitcoin market price. The findings from the research offer insights to the Bitcoin mining process suggesting that an access to extremely low electricity prices is needed to earn any profits from the Bitcoin mining activity in 2020.

This thesis analyzes behavior of customers on an e-commerce website in order to predict whether the customer is willing to buy something or is just window shopping. In addition the secondary model predicts, if the customer is going to leave the e-commerce website in next few clicks. To answer this questions different frameworks are tested. The base model used is the Logit model. The base model is compared with more sophisticated methods in machine learning - with neural networks. The best results were yielded by Recurrent neural network - the Long Short-Term Memory (LSTM). The results of the analysis confirm importance of the click stream data and calculated features that track user behavior on the e-commerce website, type of the page (product, category, information), product variance and category variance. The thesis emphasizes practical implications of this models. Two possible practical implementations are presented. The models are tested in novel ways to see how would they perform if implemented on the real e-commerce website.

In this thesis, we investigate the impact of intermittent renewable energy sources on the level and volatility of the Czech electricity spot prices dur- ing the period from 2015 to 2019. The analysis is warranted due to the obligations of the member states of the European Union to augment the share of clean energy in the gross final energy consumption by 2030. The technique applied in the empirical part concerns univariate GARCH-class models (namely, plain vanilla and exponential) which are extended with additional explanatory variables in the form of total load, solar and wind power generations. By constructing daily, peak and off-peak indices from the dataset comprised of hourly observations, we establish a comparative framework throughout the text. More specifically, this approach allows us to contrast price dynamics under the regimes of high and low demand for electricity as well as to explore the patterns of solar and wind production. The findings indicate that both Czech solar and wind power sources induce the so-called merit order effect. In contrast, once the volatility of electric- ity prices is taken into account, the examined sources of energy behave in a different manner. Owing to the daily index, while solar power generation decreases the volatility of electricity prices, the opposite...

Machine learning is becoming a part of everyday life and has an indisputable impact across large array of industries. In the financial industry, this impact lies particularly in predictive modelling. The goal of this thesis is to describe the basic principles of artificial intelligence and its subset, machine learning. The most widely used machine learning techniques are outlined both in a theoretical and a practical way. As a result, four models were assembled within the thesis. Results and limitations of each model were discussed and these models were also mutually compared based on their individual per- formance. The evaluation was executed on a real world dataset, provided by Home Credit company. Final performance of machine learning methods, measured by the KS and GINI metrics, was either very comparable or even worse than the performance of a traditional logistic regression. Still, the problem may lie in an insu cient dataset, in the improper data prepara- tion, or in inappropriately used algorithms, not necessarily in the models themselves.

Bibliographic note TRAN NGUYEN, Thai Nhat Phi. Fundamental Analysis and Stock Return: The Case Of Big Tech. Prague 2020. 102 pp. Bachelor thesis (Bc.) Charles University, Faculty of Social Sciences, Institute of Economic Studies. Thesis supervisor doc. PhDr. Ladislav Krištoufek Ph.D. Abstract Six out of the ten most valuable companies by market capitalisation are, at their core, technology companies and four of these have at some time crossed the $1 trillion market cap, which has ignited a public discussion regarding their astronomic valuations and the tech bubble. This work addresses this development, with the analysis of four companies, namely Google, Apple, Facebook and Amazon (GAFA), which have dominated their respective fields of business in the "new economy". We go beyond the stock analysis and also examine the company's fundamentals and their effect on the valuations, furthermore we fuse the insights of both analyses to offer a more comprehensive evaluation of these four companies. The results suggest that their stock value accurately portrays their market dominance and that it is deeply rooted in the companies' fundamentals which are fairly well reflected in the stock price movements. Ultimately, we find that these companies do not contribute to the tech bubble as GAFA show unparalleled financial...

1 Abstract The phenomenon of the gender pay gap has now been vastly examined for sev- eral decades and the research to date is extensive and diverse. In the thesis at hand, we accumulate, review and summarize empirical findings across the recent literature on the gender pay gap by means of meta-regression analysis. Our sample covers almost 20 years of the latest research and 35 developed coun- tries. The meta-regression results explain almost 90 percent of the study-to- study variation and suggest that the largest impact on the estimated adjusted gender pay gap results from not controlling for hours worked. Furthermore, the publication status of the original study is shown to have a large downward impact on the estimated gender pay gap. Conversely, the empirical findings suggest that a study affiliated with a certain institution promoting feminism or gender equality tend to overestimate the adjusted gender pay gap. JEL Classification J16, J31, J71 Keywords gender pay gap, discrimination Title A Meta-Analytic Approach to Gender Pay Gap: A Case of Discrimination?

The topic of our thesis is the examination of the relationships among cryptocur- rencies using Google Trends. In our thesis, we concentrated on four cryptocur- rencies, namely: Bitcoin, Litecoin, Ethereum Classic and Ethereum. We obtained the data of daily opening prices, daily trading volumes and daily Google Trends queries in order to examine the relationships among the four cryptocurrencies. Applying the Vector autoregression model and Vector error correction model, we constructed four models. The first model contains only four time series of daily prices of cryptocurrencies. The second model is the first model enriched by the respective four time series of Google Trends queries. The third model contains the four time series of daily trading volumes of the four cryptocurrencies. The fourth model is the third model enriched by the four time series of Google Trends queries of respective cryptocurrencies. Then we applied the Impulse response analysis and the Forecast error variance decomposition in order to find some relationships among the variables. We found that there is some correlation among prices, volumes and Google Trends queries containing the names of the four cryptocurrencies. According to our results acquired by the Forecast error variance decomposition, in all our models, Bitcoin has the...

In this thesis, we propose alternative ways to apartments' mass appraisal. This work enriches the current literature by combining several techniques of data extraction and price estimation. We are not aware of any similar work providing an in-depth overview of the Czech apartment market. Throughout the empirical analysis, five different methods (OLS, LASSO, decision tree, random forests, and kNN) are applied to the dataset of 15,848 classifieds. The aim of the study is to find the most accurate method of esti- mating offering prices, using structured variables as well as data extracted by text mining. We use various accuracy statistics and graphical analysis to vali- date our results. Tree-based methods, specifically the random forest algorithm, results with the highest accuracy in predicting offering prices. Additionally, text-based variables included in the model cause the reduction of errors on linear models. The last part of the analysis covers the main determinants of property value in Prague and the rest of the Czech Republic. We show that prices in Prague can be estimated with higher preciseness and with the lower number of independent variables.

The work covers the topic of private equity funds performance and attempt to identify the impact of macroeconomic conditions on the entire industry. The recent central banks' actions put a question about the impact of changes in interest rates on the private equity funds performance. With the sample of 100 observations provided by Cambridge Associates, we identified the significant negative effect of prevailing low interest rates on the growth of private equity funds performance. We further attempt to answer the question, whether private equity funds operating in post-crisis years has on average higher growth rate, however, we could not provide the answer as we failed to reject the null, neutral effect hypothesis. Additionally, with a sample of 3092 observations provided by Bloomberg, we found that the effect of cheap debt has increased on average in the postcrisis period, predicting that the private equity performance can suffer once the interest rates rises enough.

Oil market pricing is highly susceptible to geopolitical and economic events. With the rapid development of information technology, energy market can quickly get external information shocks through the Internet. This thesis examines the relationship between prices of three oil benchmarks, CBOE Crude Oil Volatility Index, and Google search queries. We built VAR model to study Granger causality and to provide impulse response analysis. Results indicate both one side and two-side causal relationship between oil-related series and most of the search queries. Out-of sample forecasting with measures of predictive accuracy and Diebold-Mariano test demonstrated that Google trends can improve short-run prediction potential only for models with WTI price and volatility index.

Cryptocurrencies are well known asset that almost everyone at least heard of. Some of these cryptocurrencies have all of its supply already in circulation, while others do not. Two biggest cryptocurrencies, Ethereum and Bitcoin, have their supply being increased every day through the process known as mining, which many never even heard of. This process and its interaction with price are the main topic of this thesis. Price and hash rate, which is good and flexible index of difficulty of mining, of Bitcoin and Ethereum are studied and it is shown, that there is an interplay among both currencies but also withing the currency itself.

The thesis attempts to determine how strategies used for forecasting and trad- ing on foreign exchange and stock markets perform when applied to cryptocur- rency markets. The approaches explored are ARIMA, VAR, MA Crossover, and Granger Causality using gold prices and S&P 500. The currencies traded are Bitcoin, Ethereum, Binance Coin, and Basic Attention Token. The models are trained on logarithmically transformed and differenced time series composed of the currencies' daily and hourly closing prices. Applying these strategies mostly leads to ambiguous results, with MA Crossover generally performing better than VAR, which in turn performs better than ARIMA. However, every strategy was moderately successful for at least one of the currencies examined. Trading on the hourly dataset was negatively influenced by sudden price jumps. ARIMA and VAR perform better in the inter-bubble periods. No significant Granger causality was found. Keywords Cryptocurrency, Trading, Bitcoin, Ethereum, Binance Coin, Basic Attention Token, ARIMA, VAR, MA Crossover, Granger Causality Title The Profitability of Standard Trading Strategies in Cryptocurrency Markets Author's e-mail miroslav.duda11@gmail.com Supervisor's e-mail ladislav.kristoufek@fsv.cuni.cz

This thesis investigates diversification benefits of Bitcoin and Ethereum. Technological innovation that made them possible, is interesting but for investors hard to grasp. The more important question is whether they should buy the digital currency or avoid it. We analyze Bitcoin and Ethereum from point of view of an investor within compatible with mean-variance (and non-mean-variance respectively) framework. Both cryptoassets are alternately added to base portfolio consisting of global indices representing American, European and Asian markets. Statistically rigorous tests suggest that Bitcoin yields ad- ded value to investors with utility function consistent with mean-variance setting. Same holds for for investors with preferences described by exponential and power utility func- tion. Ethereum shows similar results with exception of exponential utility. Performance benefits of both assets are preserved in the out-of-sample setting as size of test window reaches 28 weeks and increases. In the case of shorter test window, base assets show similar or slightly superior performance. Optimal allocation in out-of-sample framework is found by direct utility maximization with gradient based method. Keywords Bitcoin, Ethereum, digital currency, investment portfolio, diversification

Bibliographic note BUBNIAK, Peter. The impact of macroeconomic news announcements on the value and volatility of selected foreign exchange rates in EU. Prague 2019. 47 pp. Bachelor thesis (Bc) Charles University, Faculty of Social Sciences, Institute of Economic Studies. Thesis supervisor: Mgr Nicolas Fanta. Abstract This work analyzes the influence of positive and negative macroeconomic news on the value of exchange rate and volatility. We have chosen EUR/USD, EUR/CZK and USD/CZK as our exchange rates. The influence of macroeconomic news published by Czech national bank and European central bank were analysed. For our purposes were used econometric models GARCH(1,1) and EGARCH(1,1) with both Normal and Student's distribution of error terms. One of the major outcomes were the importance of macroeconomic news on value and volatility on the exhcange rates. For each exchange rate has effect different macroeconomic index. The crucial are: Consumer price index and Harmonised Index of Consumer Pirces, unemplyoment rate and PRIBOR and EURIBOR. Another conclusion was that our financial dataset displays the main nature of volatility. JEL Classification C22, E00, E52, E58, F3, F4, F31, G1, G13, G14 Key words financial market, exchange rate, ARCH model, GARCH model, volatility Authors e-mail bubniak.peter@gmail.com...

In an effort to examine role of capital markets' efficiency in transmission of monetary policy, 28 time series of market efficiency development are estimated with use of long-term memory and fractal dimension measures and a panel of 27 inflation targeting countries is constructed to run a random effect regres- sion. The cases of Czech Republic and Austria are thereafter more closely examined with use a vector-autoregressive and threshold vector-autoregressive frameworks on macroeconomic data spanning from 1996:Q3 to 2018:Q4. The evidence obtained through the conducted analyses support the hypothesis, that a more efficiently functioning capital market better contributes to monetary policy pass-through, or conversely, that high transaction costs, barriers to cap- ital market entry, or poor information availability may hinder the effects of central bank's monetary policy. JEL Classification F12, F21, F23, H25, H71, H87 Keywords capital market efficiency, inflation targeting, monetary transmission mechanism Author's e-mail teddy.krejci@gmail.com Supervisor's e-mail LK@fsv.cuni.cz

This bachelor thesis concerns itself with multiple objectives. First, to compare two apparently contradictory frameworks, namely the Log-periodic Power Law model and the Critical Slowing Down, suggested as being able to detect the end of financial bubbles. Second, to enrich current literature dedicated to the Log- periodic Power Law model with a comprehensible description of the non-linear optimization methods in one piece of work. This work, furthermore, aims to com- pare the performance and the robustness of two versions of this model. Regarding the Critical Slowing down, the correlation across the world market over time prior to a crash is investigated as an addition to two already studied indicators, 1-lag serial correlation and standard deviation of detrended fluctuations. Eventually, both the Log-periodic Power Law models were proved to be able to identify the time of the burst of the financial bubble, while the modified version of the model was found to be more proficient over the initial one in terms of computational efficiency and robustness. In the case of the Critical Slowing Down, obeying auto- correlation of residuals and cross-correlation of intermarket residuals came out to be misleading, and only variance was supported as an appropriate indicator of an imminent tumble, and it was proposed...

Unemployment rate is a crucial macroeconomic aspect for each state, which aim to have it as low as possible. However, if it is too low, many problems could arise due to a large number of job vacancies and a small number of people needed for market. As the Internet is very useful nowadays, the main aim of the thesis is to investigate the relationship between the Czech unemployment rate and job search on the Internet by users who are interested in changing jobs or are unemployed and need to find some work. Thanks to the relationship, we can conclude whether online data could improve unemployment prediction, which is needed to make effective government decisions. This thesis should also provide easier and better prediction of movements in the unemployment rate, which is inaccurate as most data sources used in economics are commonly available only after a substantial lag. The study applies data freely available on the website of Integrated Portal of the Ministry of Labour and Social Affairs, which provides statistics of unemployment rates, as well as data from portal Jobs.cz, where are information about job vacancies on the portal and response of candidates to occupied positions. The thesis uses a simple autoregressive model of the unemployment in the Czech Republic and extends it with extra variables...

The aim of this thesis is to assess the effect of ECB's communication on financial market co- movements between Italy, Spain, Germany and France using MGARCH family of models. Author addresses partially the potential problem of endogeneity of central bank communication by using Composite indicator of systemic stress and excess liquidity. The author estimates the impact of ECB's communication on correlations of government bond yield changes using daily data from 2008 to 2014. For this purpose author employs bivariate diagonal BEKK(1,1) and bivariate scalar BEKK(1,1) with surprises of macroeconomic announcements under control. The results are consistent and robust for all models, the results suggest that communication does not have statistically significant effect on financial market correlations in the Euro area. Furthermore, author defines delta functions which describe and quantify the immediate and full effect of explanatory variables on conditional correlations in bivariate diagonal BEKK(1,1) and bivariate scalar BEKK(1,1). To the best of author's knowledge this thesis is the only one in the literature which examines this effect of ECB's communication by MGARCH models. Keywords: Financial markets, central bank communication, correlation, MGARCH, BEKK Author's e-mail: milankucera1@seznam.cz...

Pairs trading is a trading strategy which tries to exploit mean-reversion among prices of certain securities. It is market-neutral and self-financing, and has been shown to produce high excess returns in historical backtests. We employ the most common distance and cointegration approaches on cryp- tocurrency data from an exchange called Binance spanning the year 2018. The strategy is mostly unprofitable under transaction costs, but certain combinations of hyperparameters can perform well. Overall, the distance method performs far better, being able to achieve 3% monthly profit even in our baseline real-life con- ditions while the cointegration method always achieves only a slight loss. We also found that increasing the sampling frequency of the data from daily to hourly brings mixed results. Moreover, since we have to reuse estimates of real-life considerations from equity markets, it is unclear if our results are truly representative of the cryp- tocurrency market. The strategy is found to be very sensitive to execution diffi- culties and transaction costs, making their determination crucially important. It is somewhat easy to get returns in excess of 5% monthly under ideal conditions, but whether this could be achieved in real trading conditions is still unclear. Keywords pairs trading,...

Ten years have passed since the emergence of Bitcoin and with it cryptocur- rencies as a new class of assets. Now, cryptocurrencies are not uncommon tool of investment and subject of academic research. This thesis focuses on investigating possible presence of weekly and monthly seasonal patterns in cryptocurrencies, namely Bitcoin, Litecoin, Ripple, Monero, Dash, Stellar and partly Ethereum, which are selected as representative sample. Insuffi- cient evidence is found for the day-of-the-week effect, the January effect is however revealed as significant by different methods in the whole sample, with cryptocurrencies generally exhibiting higher returns towards the end of the year and lowest from January to March. Examining probable causes of revealed seasonality, it is found that these are not likely to be caused by peculiar price development in 2017 and 2018, as well as the Chinese New Year or brought to the market by proposed price drivers of Bitcoin. How- ever, significant evidence for correlation of patterns followed by Bitcoin and other examined cryptocurrencies is found.