Given the volatility of the stock market and the multitude of financial variables at play, forecasting the value of stocks can be a challenging task. Nonetheless, such prediction task presents a fascinating problem to solve using machine learning. The stock market can be affected by news events, social media posts, political changes, investor emotions, and the general economy among other factors. Predicting the stock value of a company by simply using financial stock data of its price may be insufficient to give an accurate prediction. Investors often openly express their attitudes towards various stocks on social medial platforms. Hence, combining sentiment analysis from social media and the financial stock value of a company may yield more accurate predictions. This thesis proposes a method to predict the stock market using sentiment analysis and financial stock data. To estimate the sentiment in social media posts, we use an ensemble-based model that leverages Multi-Layer Perceptron (MLP), Long Short-Term Memory (LSTM), and Convolutional Neural Network (CNN) models. We use an LSTM model for the financial stock prediction. The models are trained on the AAPL, CSCO, IBM, and MSFT stocks, utilizing a combination of the financial stock data and sentiment extracted from social media posts on Twitter between the years 2015-2019. Our experimental results show that the combination of the financial and sentiment information can improve the stock market prediction performance. The proposed solution has achieved a prediction performance of 74.3%.

Diploma thesis is dedicated to purchasing strategy evaluation in ABB Brno and its optimization. Thesis analyses current status of the firm and processes, which relates to purchasing this commodity. The concept of solving includes making a new evaluation method of supplier’s non-financial data and its influence to a final evaluation of suppliers. This method is also used for real case of purchasing this commodity. Last part also defines requirements of realization and benefits for this firm.

Our objective of this thesis is to expand the microdata database of publicly available corporate information of the university by web scraping mechanism. The tool for this thesis is a web scraper that can access and concentrate information from websites utilizing a web application as an interface for client connection. In our comprehensive work we have demonstrated that the GRI text files approximately consist of 7227 companies; from the total number of companies the data is filtered with “listed” companies. Among the filtered 2252 companies some do not have income statements data. Hence, we have finally collected data of 2112 companies with 36 different sectors and 13 different countries in this thesis. The publicly available information of income statements between 2016 to 2020 have been collected by GRI of microdata department. Collecting such data from any proprietary database by web scraping may cost more than $ 24000 a year were collecting the same from the public database may cost almost nil, which we will discuss further in our thesis.In our work we are motivated to collect the financial data from the annual financial statement or financial report of the business concerns which can be used for the purpose to measure and investigate the trading costs and changes of securities, common assets, futures, cryptocurrencies, and so forth. Stock exchange, official statements and different business-related news are additionally sources of financial data that individuals will scrape. We are helping those petty investors and students who require financial statements from numerous companies for several years to verify the condition of the economy and finance concerning whether to capitalise or not, which is not possible in a conventional way; hence they use the web scraping mechanism to extract financial statements from diverse websites and make the investment decisions on further research and analysis.Here in this thesis work, we have indicated the outcome of the web scraping is to keep the extracted data in a database. The gathered data of the resulted database can be implemented for the required goal of further research, education, and other purposes with the further use of the web scraping technique.

Machine learning and neural networks have recently become very popular in a large category of domains, partly thanks to their ability to solve complex problems by finding patterns in data, but also due to an increase in computing power and data availability. Successful applications of machine learning include for example image classification, natural language processing, and product recommendation. Despite the potential upside of machine learning applied to financial data there exists relatively few articles published while the ones that do exist exhibit that there exists a potential for the tools that it provides. This thesis utilizes neural networks to price collateralized loan obligations which is a type of bond that is backed by a large pool of corporate loans, rather than being issued by a single company or government like a regular bond. The large pool of corporate loans and structure of a collateralized loan obligation makes it a good candidate for this type of research as it involves regressing a large number of variables into a final single real-valued price of the bond where the relations are not necessarily linear. The thesis establishes a relatively simple model and builds upon this using a state-of-the-art ensemble method while also exploring a volatility scaled loss function. The findings of this thesis are that artificial neural networks can price collateralized loan obligations using only their structural and loan pool data with an accuracy close to that of a human. Ensemble methods outperform non-ensemble methods and boost performance by up to 28% when only considering mean squared error while scaling the loss function with the inverse of market volatility does not boost performance. The best performing model can price a collateralized loan obligation tranche rated AAA with an average absolute error of 0.88 and an equity tranche with an average mean absolute error of 4.67. / Under de senaste åren har maskininlärning samt artificiella neurala nätverk blivit väldigt populära i många olika domäner. Detta är delvis tack vare deras förmåga att lösa komplexa problem genom att hitta mönster i data, men även tack vare en ökning i beräkningskraft samt att tillgängligheten av data har blivit bättre. Några exempel på områden där maskininlärning har applicerats med framgång är klassificering av bilder, språkteknologi samt produktrekommendationer. Trots att maskininlärning skulle kunna erbjuda en stor potentiell uppsida vid lyckad tillämpning på finansiella data finns relativt lite studier publicerade kring ämnet. De studier som däremot är publicerade visar på stora möjligheter inom området. Den här studien använder artificiella neurala nätverk för att prissätta ”collateralized loan obligations” (CLOs), som tyvärr inte har någon bra svensk översättning. En CLO utfärdar obligationer vars underliggande värde härstammar från en portfölj av företagslån, och är därmed ett finansiellt instrument. Strukturen av en CLO och dess underliggande lånportfölj ger upphov till en stor mängd data, vilket gör instrumentet till en bra kandidat för maskininlärning. Studien etablerar ett relativt enkelt neuralt nätverk som sedan används för ett jämföra med en ensemblemetod samt en modifierad loss funktion som tar höjd för volatilitet. Slutsatserna av den här studien är att neurala nätverk lyckas prissätta instrumenten näst intill lika bra som vad en människa skulle kunna göra med befintliga metoder som bygger på Monte Carlo simulering. Däremot är studiens metod inte lika beroende av antaganden som gör den befintliga metoden väldigt känslig. Vidare så bidrar ensemblemetoden som används till att minska det genomsnittliga felet i kvadrat med upp till 28%. Att ta höjd för volatilitet vid inlärning bidar inte till att minska felet.

This study investigates the prediction of firms’ health in terms of bankruptcy and non-bankruptcy based on the sentiments extracted from the earnings calls. Bankruptcy prediction has long been a critical topic in the world of accounting and finance. A firm's economic health is the current financial condition of the firm and is crucial to its stakeholders such as creditors, investors, shareholders, partners, and even customers and suppliers. Various methodologies and strategies have been proposed in research domain for predicting company bankruptcy more promptly and accurately. Conventionally, financial risk prediction has solely been based on historic financial data. However, an increasing number of finance papers also analyze textual data during the last few years. Company’s earnings calls are the key source of information to investigate the current financial condition and how the businesses are doing and what the expectations are for the next quarters. During the call, management offers an overview of recent performance and provide a guidance for the next quarter expectations. The earnings calls summary is provided by the management and can extract the CEO’s sentiments using sentiment analysis. In the last decade, Machine Learnings based techniques have been proposed to achieve accurate predictions of firms’ economic health. Even though most of these techniques work well in a limited context, on a broader perspective these techniques are unable to retrieve the true semantic from the earnings calls, which result in the lower accuracy in predicting the actual condition of firms’ economic health. Thus, state-of-the-art Machine Learnings and Deep Learnings techniques have been used in this thesis to improve accuracy in predicting the firms’ health from the earnings calls. Various machine learnings and deep learnings method have been applied on web-scraped earnings calls data-set, and the results show that LONG SHORT-TERM MEMORY (LSTM) is the best machine learnings technique as compared to the comparison set of models.

Made available in DSpace on 2016-04-26T20:30:29Z (GMT). No. of bitstreams: 1 Rafhael Wasserman.pdf: 2007952 bytes, checksum: 08ef0dea0aaba342671f03cfa8e93d38 (MD5) Previous issue date: 2010-05-25 / Conselho Nacional de Desenvolvimento Científico e Tecnológico / The scope of this study is to look into tax-related information from the moment it is seized to the moment it is used as evidence of fines and taxes levied. This study is justified due to the vulnerableness of the individual s fundamental right to privacy. The right to privacy, especially in terms of protection of financial and tax-related data, is protected by confidentiality provisions violated by amendments to the National Tax Code resulting from Supplementary Laws 104 and 105, both from January 10, 2001. Firstly, we will examine the Brazilian Revenue Service and the myriad of tools it has available to inspect the lives and activities of individuals and legal entities, from the right of scrutinizing accounting books, merchandise, files, and documents, to the right of imposing the duty to provide information on the taxpayer to financial institutions and the like. There is an undeniable tension between the forms of information gathering and the protection of confidential data. Among the kinds of confidentiality related to our topic, financial data confidentiality stands out. According to prevailing case law and scholarly opinions, financial data confidentiality can be moderated as a result of a court order. However, contrariwise, Supplementary Law 105/01 has authorized data to be directly turned over to the Revenue Service. This is a clear non-conformity with the current Brazilian constitutional system. Provided constitutional provisions are taken into account, tax-related information are undeniably subject to being transferred to the tax authorities, which, in turn, have the duty of keeping them from third parties due to the confidentiality clause. This second kind of data confidentiality protection ensures the same right to privacy by preventing said information from being disclosed to third parties. This provision was made more elastic by Supplementary Law 104/01. Likewise and for the same reasons as financial confidentiality, only a court order can break through the confidential nature of tax-related information. Tax-related data, provided they are lawfully obtained, can be employed by the tax authorities in order to produce evidence of taxes and fines levied, and issue deficiency notices. The evidence submitted by the Revenue Service shall be admitted as long as it respects the applicable constitutional and legal provisions, especially provisions related to individual rights and freedoms. We understand, differently from the current and prevailing literature, which seems to passively accept the full disclosure of tax-related information on taxpayers and third parties based on the recent Supplementary Laws , that although tax inspection fulfills the administration s revenue needs, its limits are drawn by constitutional provisions, which no other kind of legislation has the power to disregard / Este trabalho tem como escopo a análise das informações fiscais, do momento de sua apreensão à sua utilização, sobremodo como provas a lastrear a exigência de tributos e multas. Justifica-se a reflexão em razão da vulneração ao direito fundamental à privacidade dos cidadãos, tutelado por normas de sigilo de dados, em especial os sigilos financeiro e fiscal, por força das substanciais alterações ao texto do Código Tributário Nacional oriundas do advento das Leis Complementares nº 104 e 105, ambas de 10 de janeiro de 2001. Parte-se do exame da Administração Tributária e do vasto instrumental à sua disposição para fiscalizar as atividades desenvolvidas pelos particulares, desde o direito de examinar livros, mercadorias, arquivos e documentos dos sujeitos passivos, à imposição de deveres de informar a contribuintes e terceiros, como instituições financeiras e entes assemelhados. Observa-se uma inegável tensão entre essas formas de coleta de informações e o sigilo de dados. Dentre as espécies de sigilo de dados relacionadas à temática, desponta o sigilo financeiro, passível de relativização mediante decisão judicial, ao contrário do insculpido na Lei Complementar nº 105/01, que autoriza a transferência direta de dados à Fazenda Pública, em desconformidade à ordem constitucional vigente. Respeitadas as balizas constitucionais, as informações serão passíveis de comunicação às autoridades fiscais, que têm o dever de mantê-las afastadas do conhecimento alheio, por influxo do sigilo fiscal. Essa outra espécie de sigilo de dados atua na proteção do mesmo direito à privacidade, ao impedir a revelação de tais informações a terceiros, cujo regramento foi flexibilizado com a edição da Lei Complementar nº 104/01. Da mesma forma que o sigilo financeiro e pelas mesmas razões, o sigilo fiscal somente admite afastamento por meio de decisão judicial. Os dados de matiz tributário, quando licitamente produzidos, poderão ser apropriados por agentes fiscais na forma de provas a lastrear a exigência de tributos e multas, por meio da composição de atos administrativos de lançamento ou auto de infração. As provas constituídas pela Administração serão reputadas admissíveis desde que respeitadas as normas constitucionais e legais aplicáveis, mormente os direitos e garantias individuais. Entende-se contrariamente à tendência doutrinária atual, que aceita passivamente a ampla divulgação de informes fiscais relacionados a contribuintes e terceiros em decorrência da nova legislação complementar, pois a fiscalização tributária, embora indispensável à realização do interesse arrecadatório, encontra limites delineados pelo legislador constituinte, os quais não são superáveis por enunciados infraconstitucionais

Submitted by Thiago Winkler Alves (thiagowinkler@gmail.com) on 2014-09-04T13:34:50Z No. of bitstreams: 1 forecasting-daily-volatility.pdf: 885976 bytes, checksum: 30fb655def03c3f3e61bf930b3a3585b (MD5) / Approved for entry into archive by JOANA MARTORINI (joana.martorini@fgv.br) on 2014-09-04T13:44:59Z (GMT) No. of bitstreams: 1 forecasting-daily-volatility.pdf: 885976 bytes, checksum: 30fb655def03c3f3e61bf930b3a3585b (MD5) / Made available in DSpace on 2014-09-04T13:51:17Z (GMT). No. of bitstreams: 1 forecasting-daily-volatility.pdf: 885976 bytes, checksum: 30fb655def03c3f3e61bf930b3a3585b (MD5) Previous issue date: 2014-08-06 / Aiming at empirical findings, this work focuses on applying the HEAVY model for daily volatility with financial data from the Brazilian market. Quite similar to GARCH, this model seeks to harness high frequency data in order to achieve its objectives. Four variations of it were then implemented and their fit compared to GARCH equivalents, using metrics present in the literature. Results suggest that, in such a market, HEAVY does seem to specify daily volatility better, but not necessarily produces better predictions for it, what is, normally, the ultimate goal. The dataset used in this work consists of intraday trades of U.S. Dollar and Ibovespa future contracts from BM&FBovespa. / Objetivando resultados empíricos, este trabalho tem foco na eaplicação do modelo HEAVY para volatilidade diária com dados financeiros do mercado Brasileiro. Muito similar ao GARCH, este modelo busca explorar dados em alta frequência para atingir seus objetivos. Quatro variações dele foram então implementadas e seus ajustes comparadados a equivalentes GARCH, utilizando métricas presentes na literatura. Os resultados sugerem que, neste mercado, o HEAVY realmente parece especificar melhor a volatilidade diária, mas não necessariamente produz melhores previsões, o que, normalmente, é o objetivo final. A base de dados utilizada neste trabalho consite de negociações intradiárias de contratos futuros de dólares americanos e Ibovespa da BM&FBovespa.

This thesis analyzes techniques of manipulation of accounting data for the purpose of fraud. It is further looking for methods, which could be capable of detecting these manipulations and it verifies the efficiency of the procedures already in use. A theoretical part studies method of financial analysis, statistical methods, Benford's tests, fuzzy matching and technologies of machine learning. Practical part verifies the methods of financial analysis, Benford's tests, algorithms for fuzzy matching and neural networks.

Cette thèse a été réalisée au sein de l’entreprise Invivoo. L’objectif principal était de trouver des stratégies d’investissement : avoir un gain important et un risque faible. Les travaux de recherche ont été principalement portés par ce dernier point. Dans ce sens, nous avons voulu généraliser un modèle fidèle à la réalité des marchés financiers, que ce soit pour des données à basse comme à haute fréquence et, à très haute fréquence, variation par variation. / No English summary available.

Digitalization has led to tons of available customer data and possibilities for data-driven innovation. However, the data needs to be handled carefully to protect the privacy of the customers. Generative Adversarial Networks (GANs) are a promising recent development in generative modeling. They can be used to create synthetic data which facilitate analysis while ensuring that customer privacy is maintained. Prior research on GANs has shown impressive results on image data. In this thesis, we investigate the viability of using GANs within the financial industry. We investigate two state-of-the-art GAN models for synthesizing tabular data, TGAN and CTGAN, along with a simpler GAN model that we call WGAN. A comprehensive evaluation framework is developed to facilitate comparison of the synthetic datasets. The results indicate that GANs are able to generate quality synthetic datasets that preserve the statistical properties of the underlying data and enable a viable and reproducible subsequent analysis. It was however found that all of the investigated models had problems with reproducing numerical data. / Digitaliseringen har fört med sig stora mängder tillgänglig kunddata och skapat möjligheter för datadriven innovation. För att skydda kundernas integritet måste dock uppgifterna hanteras varsamt. Generativa Motstidande Nätverk (GANs) är en ny lovande utveckling inom generativ modellering. De kan användas till att syntetisera data som underlättar dataanalys samt bevarar kundernas integritet. Tidigare forskning på GANs har visat lovande resultat på bilddata. I det här examensarbetet undersöker vi gångbarheten av GANs inom finansbranchen. Vi undersöker två framstående GANs designade för att syntetisera tabelldata, TGAN och CTGAN, samt en enklare GAN modell som vi kallar för WGAN. Ett omfattande ramverk för att utvärdera syntetiska dataset utvecklas för att möjliggöra jämförelse mellan olika GANs. Resultaten indikerar att GANs klarar av att syntetisera högkvalitativa dataset som bevarar de statistiska egenskaperna hos det underliggande datat, vilket möjliggör en gångbar och reproducerbar efterföljande analys. Alla modellerna som testades uppvisade dock problem med att återskapa numerisk data.