National Repository of Grey Literature 142 records found  1 - 10nextend  jump to record: Search took 0.00 seconds. 
Forecasting Electricity Pricing in Central and Eastern Europe
Křížová, Kristýna ; Krištoufek, Ladislav (advisor) ; Baruník, Jozef (referee)
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...
Analysis of the US stock market during the COVID-19 pandemic
Tůma, Adam ; Krištoufek, Ladislav (advisor) ; Fanta, Nicolas (referee)
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.
Forex forecasting with Support vector regression and Long short-term memory recurrent neural network
Bodický, Michal ; Šíla, Jan (advisor) ; Krištoufek, Ladislav (referee)
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
Three Essays on Bank-Sourced Credit Risk Estimates
Štěpánková, Barbora ; Krištoufek, Ladislav (advisor) ; Teplý, Petr (referee) ; Seow, Hsin-Vonn (referee) ; Ansell, Jake (referee)
The aim of the thesis is to bring new insights into banks' internal credit risk estimates and their application in estimation of credit transition matrices, which are an important part of credit risk modelling with limited publicly available sources. The doctoral thesis consists of three essays that jointly analyse features of bank- sourced credit risk data and practicalities of transition matrices estimation. In the first essay, I empirically test two assumptions widely used for estimation of transition matrices: Markovian property and time homogeneity. The results indicate that internal credit risk estimates do not satisfy the two assumptions, showing evidence of both path-dependency and time heterogeneity even within a period of economic expansion. Contradicting previous findings based on data from credit rating agencies, banks tend to revert their past rating actions. The second essay analyses the extent to which transition matrices depend on the characteristics of the underlying overlapping bank-sourced credit risk datasets and the aggregation method. It outlines that the choice of aggregation approach has a substantial effect on credit risk model results. I also show that bank-sourced transition matrices are more dynamic than those produced by credit rating agencies and introduce industry-specific...
Fractality of stock markets: a comparative study
Krištoufek, Ladislav ; Baruník, Jozef (advisor)
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.
Liquidity and Predictability of Cryptoassets
Mjartanová, Viktória ; Krištoufek, Ladislav (advisor) ; Čech, František (referee)
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...
Using spin model to determine FTTx connectivity market potential in the Czech Republic
Munduch, Pavel ; Krištoufek, Ladislav (advisor) ; Červinka, Michal (referee)
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
Multi-horizon equity returns predictability via machine learning
Nechvátalová, Lenka ; Baruník, Jozef (advisor) ; Krištoufek, Ladislav (referee)
We examine the predictability of expected stock returns across horizons using machine learning. We use neural networks, and gradient boosted regression trees on the U.S. and international equity datasets. We find that predictabil- ity of returns using neural networks models decreases with longer forecasting horizon. We also document the profitability of long-short portfolios, which were created using predictions of cumulative returns at various horizons, be- fore and after accounting for transaction costs. There is a trade-off between higher transaction costs connected to frequent rebalancing and greater returns on shorter horizons. However, we show that increasing the forecasting hori- zon while matching the rebalancing period increases risk-adjusted returns after transaction cost for the U.S. We combine predictions of expected returns at multiple horizons using double-sorting and buy/hold spread, a turnover reduc- ing strategy. Using double sorts significantly increases profitability on the U.S. sample. Buy/hold spread portfolios have better risk-adjusted profitability in the U.S. JEL Classification G11, G12, G15, C55 Keywords Machine learning, asset pricing, horizon pre- dictability, anomalies Title Multi-horizon equity returns predictability via machine learning
A Meta-Analytic Approach to Gender Pay Gap: A Case of Discrimination?
Konstantinova, Elizaveta ; Krištoufek, Ladislav (advisor) ; Pertold-Gebicka, Barbara (referee)
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?

National Repository of Grey Literature : 142 records found   1 - 10nextend  jump to record:
See also: similar author names
2 Krištoufek, L.
Interested in being notified about new results for this query?
Subscribe to the RSS feed.