|
|
Stock Trading Using a Deep Reinforcement Learning and Text Analysis
Benk, Dominik ; Baruník, Jozef (advisor) ; Vácha, Lukáš (referee)
The thesis focuses on exploiting imperfections on the stock market by utilizing state-of-the-art learning methods and applying them to algorithmic trading. The automated decisions are expected to have the capability of outperforming professional traders by considering much more information, reacting almost instantly and being unaffected by emotions. As an alternative to traditional supervised learning, the proposed model of reinforcement learning employs a principle of trial-and-error, which is essential for learning behaviours of all organisms. In the context of stocks, this allows to consider the involved uncer- tainty and therefore more precisely estimate the long-run returns. To collect the most relevant information for each trading decision, additionally to tech- nical indicators the models build on investor's opinion - financial sentiment. This is derived from two textual sources, news and social media, and the main goal is to compare their relative contribution to trading. Models are applied to 11 different stocks and later combined into portfolio for greater robustness of results. The textual analysis proves to be important for the learning process, especially in case of stocks with good media coverage. The Twitter is found to provide more valuable information compared to news, but their...
|
|
|
Momentum trading strategy performance before, during, and after the COVID-19 crisis
Řeřicha, Dávid ; Fanta, Nicolas (advisor) ; Vácha, Lukáš (referee)
This thesis investigates the well-known momentum trading strategy from January 2013 to May 2022 on the US stock market. The goal of this thesis is to examine whether the phenomenal momentum anomalies occurred during COVID-19 crisis. The main part is addressed to the creation of momentum portfolios from the whole US stock market using daily data from 500 firms in the S&P 500 index and additional 11 sectoral momentum portfolios. Results confirm the power of momentum portfolios as the past winners accumulated the highest returns over the whole observed period and clearly outperformed the market. Focusing closely on COVID- 19 period we observed past losers outperforming past winners, which confirms another momentum anomaly on the US stock market. Therefore, this thesis referred to the Carhart Four - Factor Model model that is based on the Fama-French Three - Factor model with additional momentum factor. Unfortunately, results indicate no statistically significant power to explain the momentum behaviour during COVID-19 crisis.
|
|
| |
|
|
Price gaps in the stock market
Vosmanský, Jakub ; Krištoufek, Ladislav (advisor) ; Vácha, Lukáš (referee)
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
|
|
|
Can Machines Explain Stock Returns?
Chalupová, Karolína ; Baruník, Jozef (advisor) ; Vácha, Lukáš (referee)
Can Machines Explain Stock Returns? Thesis Abstract Karolína Chalupová January 5, 2021 Recent research shows that neural networks predict stock returns better than any other model. The networks' mathematically complicated nature is both their advantage, enabling to uncover complex patterns, and their curse, making them less readily interpretable, which obscures their strengths and weaknesses and complicates their usage. This thesis is one of the first attempts at overcoming this curse in the domain of stock returns prediction. Using some of the recently developed machine learning interpretability methods, it explains the networks' superior return forecasts. This gives new answers to the long- standing question of which variables explain differences in stock returns and clarifies the unparalleled ability of networks to identify future winners and losers among the stocks in the market. Building on 50 years of asset pricing research, this thesis is likely the first to uncover whether neural networks support the economic mechanisms proposed by the literature. To a finance practitioner, the thesis offers the transparency of decomposing any prediction into its drivers, while maintaining a state-of-the-art profitability in terms of Sharpe ratio. Additionally, a novel metric is proposed that is particularly suited...
|
|
|
Low Interest Rates and Asset Price Fluctuations: Empirical Evidence
Ali, Bano ; Horváth, Roman (advisor) ; Vácha, Lukáš (referee)
The thesis focuses on estimating the effect of expansionary monetary policy concerning asset prices, specifically house and stock prices as they are of pri- mary importance in financial markets. A structural vector autoregressive model is used including data for the Euro Area, the United Kingdom, and the United States from 2007 to 2017. Moreover, instead of short-term nominal interest rate, the shadow policy rate is used to measure the stance of both conventional and unconventional monetary policy. It is useful when policy rates of central banks are at or near zero as it neglects the zero-lower bound. Using both impulse response functions and forecast error variance decomposition, results suggest that higher interest rates are indeed associated with lower asset prices. That is confirmed by including two different estimates of shadow rates into the model and observing the effect for two specific types of assets. More precisely, house prices react almost immediately showing the most substantial decrease for the United Kingdom, while stock prices slightly increase at first and de- crease afterward with similar size of the effect for all areas under consideration. Finally, the discussion of how the monetary authority should react to asset price fluctuations is provided, summarizing the vast amount of literature...
|
|
|
Time-scale analysis of sovereign bonds market co-movement in the EU
Šmolík, Filip ; Vácha, Lukáš (advisor) ; Krištoufek, Ladislav (referee)
The thesis analyses co-movement of 10Y sovereign bond yields of 11 EU mem- bers (Greece, Spain, Portugal, Italy, France, Germany, Netherlands, Great Britain, Belgium, Sweden and Denmark) divided into the three groups (the Core of the Eurozone, the Periphery of the Eurozone, the states outside the Eurozone). In the center of attention are changes of co-movement in the crisis period, especially near the two significant dates - the fall of Lehman Brothers (15.9.2008) and the day, when increase of Greek public deficit was announced (20.10.2009). Main contribution of the thesis is usage of alternative methodol- ogy - wavelet transformation. It allows to research how co-movement changes across scales (frequencies) and through time. Wavelet coherence is used as well as wavelet bivariate and multiple correlation. The thesis brings three main findings: (1) co-movement significantly decreased in the crisis period, but the results differ in the groups, (2) co-movement significantly differs across scales, but its heterogeneity decreased in the crisis period, (3) near to the examined dates sharp and significant decrease of wavelet correlation was observable across lower scales in some states. JEL Classification C32, C49, C58, H63 Keywords Co-movement, Wavelet Transformation, Sovereign Debt Crisis, Sovereign Bond Yields,...
|
|
|
Frequency Connectedness of Financial, Commodity, and Forex Markets
Šoleová, Juliána ; Baruník, Jozef (advisor) ; Vácha, Lukáš (referee)
This Thesis is dedicated to the variance decompositions from the VAR model un- der the Diebold, Yilmaz (2012) methodology combined with the Baruník, Křehlík (2017) method of frequencies that was used to create traditional and directional spillover tables to be compared under different frequencies. Diverse markets vari- ables were used for the analysis during the period 1/6/1999 to 29/6/2018. The S&P 500 Index represented the financial markets, EUR/USD and YEN/USD rep- resented the Forex markets, and eight types of commodities: Crude Oil, Natural Gas, Gasoline, and Propane represented energy commodities and Corn, Coffee, Wheat, and Soybeans represented food commodities. This analysis contribute to understanding of the dynamic frequency connectedness in case of a differentiated system of markets. The main finding was the strongest short-frequency reaction to shocks in case of all variables, which is opposite behavior than usually observed in banking sector frequency dynamics analyses. JEL Classication: F12, F21, F23, H25, H71, H87 Keywords: connectedness, financial market, forex market, commodity market, systemic risk, spillovers, frequency analysis Author's e-mail: 93414233@fsv.cuni.cz Supervisor's e-mail: barunik@fsv.cuni.cz
|
|
|
Modeling financial markets using heterogenous agent models
Benčík, Daniel ; Vácha, Lukáš (advisor) ; Baruník, Jozef (referee)
This thesis deals with the application of heterogeneous agent models (HAM) in the area of financial markets. In the first part, we introduce the concept of HAMs, review examples of several earlier models in order to provide the reader with a general picture of applications of HAMs in finance. Subsequently, we move on to describe the original model developed by Brock, Hommes (1998) and continue by describing modifications proposed by Barunik, Vacha and Vosvrda (2009). Next, we move to the analysis of the modified model's behavior, including its ability to simulate stylized facts observed in real financial markets. In the last part of this work, we provide descriptions of our simulation/experimental setups and conclude by summarizing the results of these. We finish this thesis by suggesting possible future research topics regarding the investigated model that might shed more light on its behavior and thus hopefully enhance our understanding of how real financial markets operate.
|
|
|
Financial markets modeling - experimental and agent based approach
Štefanová, Hana ; Vácha, Lukáš (advisor) ; Korbel, Václav (referee)
Tato práce se zabývá problémem modelování finančních trhů. K modelování používáme dva přístupy: simultánní a experimentální. Nejprve představíme agentní modelování a experimentální ekonomii. Poté vysvětlíme silné a slabé stránky těchto přístupů a ukážeme jejich společný přínos v oblasti modelování finančních trhů. Aby čtenář získal komplexnější představu o celé problematice, uvedeme několik modelů používajících kombinovanou metodologii. Následně představíme model dvojité aukce, jehož autory jsou Gode a Sunder (1993). Naši práci zakončíme výsledky experimentu, který jsme sami provedli, a jehož základní myšlenkou je právě práce od Goda a Sundera.
|
guest ::
RSS feed.