Moderne statistische und ökonometrische Methoden behandeln erfolgreich stilisierte Fakten auf den Finanzmärkten. Die vorgestellten Techniken erstreben die Dynamik von Finanzmarktdaten genauer als traditionelle Ansätze zu verstehen. Wirtschaftliche und finanzielle Vorteile sind erzielbar. Die Ergebnisse werden hier in praktischen Beispielen ausgewertet, die sich vor allem auf die Prognose von Finanzmarktdaten fokussieren. Unsere Anwendungen umfassen: (i) die Modellierung und die Vorhersage des Liquiditätsangebotes, (ii) die Lokalisierung des ’Multiplicative Error Model’ und (iii) die Erbringung von Evidenz für den empirischen Zustandsfaktorparadox über Landern. / Modern methods in statistics and econometrics successfully deal with stylized facts observed on financial markets. The presented techniques aim to understand the dynamics of financial market data more accurate than traditional approaches. Economic and financial benefits are achievable. The results are here evaluated in practical examples that mainly focus on forecasting of financial data. Our applications include: (i) modelling and forecasting of liquidity supply, (ii) localizing multiplicative error models and (iii) providing evidence for the empirical pricing kernel paradox across countries.

Este trabalho possui um duplo objetivo: i) estudar os fatos estilizados do livro de ofertas dos papéis negociados na Bolsa de Valores de São Paulo (BOVESPA), assim como dos retornos engendrados pela dinâmica do livro de ofertas e ii) desenvolver um modelo de livro de ofertas baseado em agentes com o propósito de reproduzir tais fatos estilizados. Trabalhou-se com dados de junho/2006 a janeiro/2009 de uma amostra formada pelos vinte papéis mais negociados da BOVESPA. Os resultados empíricos corroboraram alguns fatos estilizados observados no estudo de papéis de outros países, mas refutaram outros. O modelo baseado em agentes conseguiu emular satisfatoriamente os fatos estilizados relacionados aos retornos, mas em se tratando da reprodução dos fatos estilizados do livro de ofertas o modelo foi menos eficaz. / This study has two aims: i) analyze the stylized facts of the order book of stocks traded in the São Paulo Stock Exchange (BOVESPA), as well as of the returns engendered by the order book dynamics and ii) develop an order book agent-based model able to reproduce such stylized facts. It was used data from June 2006 to January 2009 regarding a sample composed by the twenty most traded stocks in BOVESPA. The empirical results corroborated some stylized facts observed in stocks of other countries, but refuted others. The agent-based model successfully emulated the stylized facts concerning the returns; however, the model was less efficient in reproducing the stylized facts of the order book.

Esta Tese explora o Processo de Cox quando sua intensidade pertence a uma família de difusões afim. A forma da funçâo densidade de Probabilidade do Processo de Cox é obtida quando a intensidade é descrita por uma difusão fim d-dimensional arbitrária. Analisa-se também o acoplamento e convergência para o Processo de Cox com intensidade afim. Para ilustrar assume-se que a intensidade do Processo é governada por uma difusão de Feller e resultados mais detalhados são obtidos. Adicionalmente, os parâmetros da intensidade do Processo são estimados por meio do Filtro de Kalman conjugado com o estimador de Quase-Máxima Verossimilhança. / This Thesis deals with the Cox Process when its intensity belongs to a family of affine diffusions. The form of the probability density function of the Cox process is obtained when the density is described by an arbitrary d-dimensional affine diffusion. Coupling and convergence results are also addressed for a general Cox process with affine intensity. We adopted the Feller diffusion for driving the underlying intensity of the Cox Process to illustrate our results. Additionally the parameters of the underlying intensity processes are estimated by means of the Kalman Filter in conjunction with Quasi-Maximum Likelihood estimation.

The present work discuss the continuous double auction mechanisms and the order book models. After a brief introduction to selected models, a general model of the the continuous double auction from the thesis title is described. Further, a structure of british market data is given as well as an approach to them. Based on these data the validity of Smith Farmer's model and Cont Stoikov's model is tested in the context of general model by linear regression. Finally, based on the previous results, the own order book model is suggested and its validity tested.

Este trabalho possui um duplo objetivo: i) estudar os fatos estilizados do livro de ofertas dos papéis negociados na Bolsa de Valores de São Paulo (BOVESPA), assim como dos retornos engendrados pela dinâmica do livro de ofertas e ii) desenvolver um modelo de livro de ofertas baseado em agentes com o propósito de reproduzir tais fatos estilizados. Trabalhou-se com dados de junho/2006 a janeiro/2009 de uma amostra formada pelos vinte papéis mais negociados da BOVESPA. Os resultados empíricos corroboraram alguns fatos estilizados observados no estudo de papéis de outros países, mas refutaram outros. O modelo baseado em agentes conseguiu emular satisfatoriamente os fatos estilizados relacionados aos retornos, mas em se tratando da reprodução dos fatos estilizados do livro de ofertas o modelo foi menos eficaz. / This study has two aims: i) analyze the stylized facts of the order book of stocks traded in the São Paulo Stock Exchange (BOVESPA), as well as of the returns engendered by the order book dynamics and ii) develop an order book agent-based model able to reproduce such stylized facts. It was used data from June 2006 to January 2009 regarding a sample composed by the twenty most traded stocks in BOVESPA. The empirical results corroborated some stylized facts observed in stocks of other countries, but refuted others. The agent-based model successfully emulated the stylized facts concerning the returns; however, the model was less efficient in reproducing the stylized facts of the order book.

This paper provides microstructure models of order-driven market to analyze the dynamic dependencies of order strategy, price formation and order book information. This study gradually derives three models to shed light on those dynamic dependencies: risk-neutral order-submission model, risk-averse order-submission model and revision order-submission model based on order book information. Those inferences support that the order-driven market dynamically adjusts the bid/ask at any moment to generate enough price improvement return in order to cover the fluctuations of the adverse selection risk and the non-execution risk faced by limit order submitters of both side. In risk-neutral order-submission model, the model anatomizes adverse selection cost and bid-ask spread under risk-neutral preference of order submitters. This study finds that adverse selection cost comprises three components: arrival probability of informed traders, execution probability of setting price of limit order, cost-to-benefit ratio of investment. In risk-averse order-submission model, the model analyzes the optimal order-submission behavior of risk-averse uninformed traders. This study finds that the asset volatility is the key determinant of the adverse selection risk and the non-execution risk, and thereby the bid-ask spread is associated with the asset volatility. The novelty approach of this model could connect both previous risk-neutral models of Handa, Schwartz and Tiwari (2003) and Foucault (1999), which are the special cases of the reduced form of this model. In revision order-submission model, the model analyzes adverse selection costs and price formation of bid-ask spread, dynamically adjusted by previous state of limit order book in an electronic limit order market. Using order book data from the Taiwan Stock Exchange, the empirical analysis corroborates the following findings: (1) the state of the limit order book significantly affects subsequent order aggressiveness; (2) adverse selection cost and spread are negatively associated with the precision of order book information; (3) information effects of limit order book on the bid-ask spread provide strong support for the model.

Cette thèse étudie quelques aspects de la modélisation stochastique des carnets d'ordres. Nous analysons dans la première partie un modèle dans lequel les temps d'arrivées des ordres sont Poissoniens indépendants. Nous démontrons que le carnet d'ordres est stable (au sens des chaines de Markov) et qu'il converge vers sa distribution stationnaire exponentiellement vite. Nous en déduisons que le prix engendré dans ce cadre converge vers un mouvement Brownien aux grandes échelles de temps. Nous illustrons les résultats numériquement et les comparons aux données de marché en soulignant les succès du modèle et ses limites. Dans une deuxième partie, nous généralisons les résultats à un cadre où les temps d'arrivés sont régis par des processus auto et mutuellement existants, moyennant des hypothèses sur la mémoire de ces processus. La dernière partie est plus appliquée et traite de l'identification d'un modèle réaliste multivarié à partir des flux des ordres. Nous détaillons deux approches : la première par maximisation de la vraisemblance et la seconde à partir de la densité de covariance, et réussissons à avoir une concordance remarquable avec les données. Nous appliquons le modèle ainsi estimé à deux problèmes concrets de trading algorithmique, à savoir la mesure de la probabilité d'exécution et le coût d'un ordre limite. / This thesis presents some aspects of stochastic order book modelling. In the first part, we analyze a model in which order arrivals are independent Poisson. We show that the order book is stable (in the sense of Markov chains) and that it converges to its stationary state exponentially fast. We deduce that the price generated in this setting converges to a Brownian motion at large time scales. We illustrate the results numerically and compare them to market data. In the second part, we generalize the results to a setting in which arrival times are governed by self and mutually existing processes. The last part is more applied and deals with the identification of a realistic multivariate model from the order flow. We describe two approaches: the first based on maximum likelihood estimation and the second on the covariance density function, and obtain a remarkable agreement with the data. We apply the estimated model to two specific algorithmic trading problems, namely the measurement of the execution probability of a limit order and its cost.

Esta Tese explora o Processo de Cox quando sua intensidade pertence a uma família de difusões afim. A forma da funçâo densidade de Probabilidade do Processo de Cox é obtida quando a intensidade é descrita por uma difusão fim d-dimensional arbitrária. Analisa-se também o acoplamento e convergência para o Processo de Cox com intensidade afim. Para ilustrar assume-se que a intensidade do Processo é governada por uma difusão de Feller e resultados mais detalhados são obtidos. Adicionalmente, os parâmetros da intensidade do Processo são estimados por meio do Filtro de Kalman conjugado com o estimador de Quase-Máxima Verossimilhança. / This Thesis deals with the Cox Process when its intensity belongs to a family of affine diffusions. The form of the probability density function of the Cox process is obtained when the density is described by an arbitrary d-dimensional affine diffusion. Coupling and convergence results are also addressed for a general Cox process with affine intensity. We adopted the Feller diffusion for driving the underlying intensity of the Cox Process to illustrate our results. Additionally the parameters of the underlying intensity processes are estimated by means of the Kalman Filter in conjunction with Quasi-Maximum Likelihood estimation.

This diploma thesis deals with the market-making strategy's profitability analysis, tested on simulation of central order book. The theoretical part describes how the market maker quotes the price of supply and demand and mathematically proves under which circumstances this strategy is profitable. The practical part introduces a simulation of the central order book. The advantage of simulating the entire order book is that we have information about the number of market participants and quotes at any given time. It also introduces a fictitious market maker quoting the price of supply and demand at any given moment, the price being determined by the price of the previous time step. The order book is simulated in three different settings - random walk, mean-reversion and leptokurtic distribution, and it is shown that the expected profitability of the market-maker strategy is positive in all three cases.

The intention of this thesis is to provide a primitive mathematical model for a financial market in which tradings affect the asset prices. Currently, the idea of a price-volume relationship is typically used in the form of empirical models for specific cases. Among the theoretical models that have been used in stock markets, few included the volume parameter. The thesis provides a general theoretical model with the volume parameter for the intention of a broader use. The core of the model is the correlation between trading volume and stock price, indicating that volume should be a function of the stock price and time. This function between price and time was made visible by the use of the trading volume process, also known as the Limit Order book. The development of this model may be of some use to investors, who could build their wealth process based on the dynamics of the process found through a Limit Order Book. This wealth process can help them build an optimal trading strategy design.