Orientador: Luiz Koodi Hotta / Dissertação (mestrado) - Universidade Estadual de Campinas, Instituto de Matemática, Estatística e Computação Científica / Made available in DSpace on 2018-08-20T22:42:22Z (GMT). No. of bitstreams: 1 TruciosMaza_CarlosCesar_M.pdf: 13820849 bytes, checksum: 0cc000af0d7cb7cb6ee6c05ef9f3afbd (MD5) Previous issue date: 2012 / Resumo: Mercados financeiros têm mostrado um grande interesse em intervalos de previsão como uma medida de incerteza. Além das previsões do nível, a previsão da volatilidade é importante em várias aplicações em finanças. O modelo GARCH tem sido bastante utilizado na modelagem da volatilidade. A partir deste modelo, outros modelos foram propostos para incorporar outros fatos estilizados, como o efeito de alavancagem. Neste sentido, temos os modelos EGARCH e GJR-GARCH. Os métodos tradicionais de construção de intervalos de previsão para séries temporais geralmente assumem que os parâmetros do modelo são conhecidos e os erros normais. Quando estas suposições não são verdadeiras, o que costuma acontecer na prática, o intervalo de previsão obtido tenderá a ter uma cobertura abaixo da nominal. Nesta dissertação propomos uma adaptação do algoritmo PRR (Pascual, Romo e Ruiz) desenvolvido para obter intervalos de previsão em modelos GARCH, para obter intervalos de previsão em modelos EGARCH e GJR-GARCH. As adaptações feitas são analisadas através de experimentos Monte Carlo e verifica-se que tiveram bom desempenho apresentando valores da cobertura estimada próximos da cobertura nominal. As adaptações propostas assim como o algoritmo PRR são aplicadas para obter intervalos de previsão dos retornos e das volatilidades para a série de retornos da Ibovespa e para a série NYSE COMPOSITE(DJ) da bolsa de valores de Nova Iorque, obtendo em ambos os casos resultados satisfatórios / Abstract: Financial Markets have shown a big interest in forecast intervals (prediction intervals) as a uncertain measure. Besides the level prediction, the prediction of the volatility is very important in many financial applications. The GARCH model, has been very used in volatility modeling. From this model, other have been proposed to incorporate other stylized facts, such as the leverage effect. In this sense, we have the EGARCH and GJR-GARCH models. Traditional methods for constructing predictions intervals for time series generally assume that the model parameters are known and the erros are normal. When these assumptions are not true, that it is very often in practice, the obtained prediction interval, will tend to have a cover under the nominal. In this theses we propose an adaptation of the PRR (Pascual, Romo and Ruiz) algorithm developed to obtain prediction intervals in GARCH models, to obtain prediction intervals in EGARCH and GJR-GARCH models. These adaptations are analized through Monte-Carlo experiments and It was verified that they have a good performance showing estimated cover values close to the nominal cover. The proposed adaptations, such as the PRR algorithm are applied to obtain prediction intervals from the returns and volatilities for the Ibovespa return series and for the New York Stock Exchange NYSE COMPOSITE(DJ) series, obtaining satisfactory results in both cases / Mestrado / Estatistica / Mestre em Estatística

[pt] O bootstrap de B. Efron, que não poderia ser imaginado sem os computadores de hoje, pode resolver vários problemas livre da suposição de Gaussianidade para os dados. Este trabalho tem o objetivo de apresentar essa técnica computacionalmente intensiva no contexto de Séries temporais - Metodologia Box and Jenkins. Como se sabe essa Metodologia possui alguns resultados assintóticos. Então, na fase da identificação da estrutura do modelo, pode apresentar problemas em regiões do espaço paramétrico aqui determinadas,. O bootstrap é proposto como opção e um estudo de simulação, comparativo, é apresentado. Constrói- se a distribuição bootstrap da autocorrelação e autocorrelação parcial, amostrais, e ainda a distribuição bootstrap do estimador de MQNL dos coeficientes de modelos ARMA (p, q). consequentemente, fica disponí­vel medida não- paramétrica da precisão da estimativa. O estudo de simulação que aborda o estimador de MQNL dos coeficientes enfoca, basicamente, a região de fronteira da estacionariedade e inversibilidade. / [en] The bootstrap of B. Efron, what should not be imagined without fast andcheaper computation, can solve several problems free from assumption that the data conform to a bell-shaped curve. This work has the aim to present this computer-intensive technics in the context of Time Series - Box and Jenkins´s Methodology. As we know this methodology own some asymptotic results. Then in the identification stage of the structure of the model it may present some troubles on regions of the parametric space, as we show later on the bootstrap is proposed as an aption and a comparative simulation study is pointed out. We build up the bootstrap distribution of the sample autocorrelation and sample partial autocorrelation, and yet a bootstrap distribution to the non-linear LS estimator of the coefficients to the ARMA (p,q) model. As a consequence we get the non- parametric measure of the accuracy of the estimates. The study of simulation wich takes into account the non-linear LS estimato to the coefficients, actually focalize the borden of the stationarity and invertibility region.

Une classe générale de statistiques de rangs basées sur la fonction caractéristique est introduite afin de tester l'hypothèse composite d'appartenance à une famille de copules multidimensionnelles. Ces statistiques d'adéquation sont définies comme des distances fonctionnelles de type L_2 pondérées entre une version non paramétrique et une version semi-paramétrique de la fonction caractéristique que l'on peut associer à une copule. Il est démontré que ces statistiques de test se comportent asymptotiquement comme des V-statistiques dégénérées d'ordre quatre et que leurs lois limites s'expriment en termes de sommes pondérées de variables khi-deux indépendantes. La convergence des tests sous des alternatives générales est établie, de même que la validité du bootstrap paramétrique pour le calcul de valeurs critiques. Le comportement des nouveaux tests sous des tailles d'échantillons faibles et modérées est étudié à l'aide de simulations et est comparé à celui d'un test concurrent fondé sur la copule empirique. La méthodologie est finalement illustrée sur un jeu de données à plusieurs dimensions.

Cette these examine deux aspects des theories conformes des champs (TCC) en d dimensions.Sa premiere parti est dediee aux blocs conformes, des fonctions speciales qui contribuent au developpement en ondes partielles des fonctions a quatre points dans les TCC. On montre que ces blocs admettent un developpement en coordonnees polaires dont les coecients se calculent par une recurrence. Les blocs conformes sont naturellement denis sur le plan complexe : on considere alors leur restriction a l'axe r eel, an de montrer qu'ils obeissent une equation dierentielle sur ce domaine, ce qui mene a un algorithme ecace pour calculer les blocs conformes et leurs derivees pour tout d. Quelques applications au programme de bootstrap sont developpees. La seconde partie de cette these examine les perturbations d'une TCC par des operateurs pertinents. On etudie de tels ots du groupe de renormalisation en utilisant la Methode de Troncature Conforme (MTC) de Yurov et Zamolodchikov, une methode numerique qui permet de faire des calculs non-perturbatifs en theorie quantique des champs. Deux theories derentes sont considerees : le boson libre avec un terme de masse, et la theorie 4. Pour le dernier cas, les resultats de la MTC mettent en evidence la brisure de symetrie Z2. Finalement, on developpe une methode pour reduire les erreurs de troncature en ajoutant des contre-termes a l'action \nue" de la MTC, suivant des travaux anterieurs en d = 2 dimensions. / This thesis investigates two aspects of Conformal Field Theories (CFTs) in d dimensions. Its rst part is devoted to conformal blocks, special functions that arise in the partial wave expansion of CFT four-point functions. We prove that these conformal blocks admit an expansion in terms of polar coordinates and show that the expansion coecients are determined by recursion relations. Conformal blocks are naturally dened on the complex plane: we study their restriction to the real line, and show that they obey a fourth-order dierential equation there. This ODE can be used to eciently compute conformal blocks and their derivatives in general d. Several applications to the conformal bootstrap program are mentioned. The second half of this thesis investigates RG ows that are dened by perturbing a CFT by a number of relevant operators. We study such ows using the Truncated Conformal Space Approach (TCSA) of Yurov and Zamolodchikov, a numerical method that allows for controlled computations in strongly coupled QFTs. Two dierent RG ows are considered: the free scalar feld deformed by a mass term, and 4 theory. The former is used as a benchmark, in order to compare numerical TCSA results to exact predictions. TCSA results for 4 theory display spontaneous Z2 symmetry breaking at strong coupling: we study the spectrum of this theory both in the Z2-broken and preserved phase, and we compare the critical exponents governing the phase transition to known values. In a separate chapter, we show how truncation errors can be reduced by adding suitable counterterms to the bare TCSA action, following earlier work in d = 2 dimensions.

Finding a well-predicting model is one of the main goals of regression analysis. However, to evaluate a model's prediction abilities, it is a normal practice to use criteria which either do not serve this purpose, or criteria of insufficient reliability. As an alternative, there are relatively new methods which use repeated simulations for estimating an appropriate loss function -- prediction error. Cross-validation and bootstrap belong to this category. This thesis describes how to utilize these methods in order to select a regression model that best predicts new values of the response variable.

La symétrie conforme impose de très fortes contraintes sur les théories quantiques des champs. En deux dimensions, l’algèbre des symétries conformes est infinie, et les théories conformes bidimensionnelles peuvent être complètement résolubles, dans le sens où toutes leurs fonctions de corrélation peuvent être calculées. Ces théories ont un grand domaine d'application, de la théorie des cordes jusqu'aux systèmes critiques en physique statistique, et elles ont été largement étudiées pendant les dernières décennies.Dans cette thèse nous étudions les théories conformes bidimensionnelles dont l’algèbre de symétrie est celle de Virasoro, en suivant l'approche connue sous le nom de bootstrap conforme. Sous l'hypothèse de l'existence de champs dégénérés, nous généralisons le bootstrap conforme analytique aux théories avec des spectres non-diagonaux. Nous écrivons les équations qui déterminent les constantes de structure, et nous trouvons des solutions explicites en termes de fonctions spéciales. Nous validons ces résultats en faisant des calculs numériques des fonctions de corrélation à quatre points dans des modèles minimaux diagonaux et non-diagonaux, et en vérifiant que la symétrie de croisement est respectée.En outre, nous construisons une proposition pour une famille de théories conformes non-diagonales et non-rationnelles pour toute charge centrale telle que Re(c) < 13. Cette proposition est motivée par les limites des spectres des modèles minimaux de la série D. Nous réalisons des calculs numériques des fonctions à quatre points dans ces théories, et nous trouvons qu'elles obéissent à la symétrie de croisement. Ces théories peuvent être interprétées comme des extensions non-diagonales de la théorie de Liouville. / Conformal symmetry imposes very strong constraints on quantum field theories. In two dimensions, the conformal symmetry algebra is infinite-dimensional, and two-dimensional conformal field theories can be completely solvable, in the sense that all their correlation functions may be computed. These theories have an ample range of applications, from string theory to critical phenomena in statistical physics, and they have been widely studied during the last decades.In this thesis we study two-dimensional conformal field theories with Virasoro algebra symmetry, following the conformal bootstrap approach. Under the assumption that degenerate fields exist, we provide an extension of the analytic conformal bootstrap method to theories with non-diagonal spectrums. We write the equations that determine structure constants, and find explicit solutions in terms of special functions. We validate this results by numerically computing four-point functions in diagonal and non-diagonal minimal models, and verifying that crossing symmetry is satisfied.In addition, we build a proposal for a family of non-diagonal, non-rational conformal field theories for any central charges such that Re(c) < 13. This proposal is motivated by taking limits of the spectrum of D-series minimal models. We perform numerical computations of four-point functions in these theories, and find that they satisfy crossing symmetry. These theories may be understood as non-diagonal extensions of Liouville theory.

Point estimates of the Standard Chain Ladder method (CLM) and of the more complex Munich Chain Ladder method (MCL) are compared to real data on 38 different datasets in order to evaluate if MCL produces better predictions on average with a dataset from an arbitrary insurance portfolio. MCL is also examined to determine if the future paid and incurred claims converge as time progresses. A bootstrap model based on MCL (BMCL) is examined in order to evaluate its possibility to estimate the probability density function (PDF) of future claims and observable claim development results (OCDR). The results show that the paid and incurred predictions by MCL converge. The results also show that when considering all datasets MCL produce on average better estimations than CLM with paid data but no improvement can be seen with incurred data. Further the results show that by considering a subset of datasets which fulfil certain criteria, or by only considering accident years after 1999 the percentage of datasets in which MCL produce superior estimations increases. When examining BMCL one finds that it can produce estimated PDFs of ultimate reserves and OCDRs, however the mean of estimate of ultimate reserves does not converge to the MCL estimates nor do the mean of the OCDRs converge to zero. In order to get the right convergence the estimated OCDR PDFs are centered and the mean of the BMCL estimated ultimate reserve is set to the MCL estimate by multiplication. / Punktskattningar gjorda med Standard Chain Ladder (CLM) och den mer komplexa Munich Chain Ladder-metoden (MCL) jämförs med verklig data för 38 olika dataset för att evaluera om MCL ger bättre prediktioner i genomsnitt än CLM för en godtycklig försäkringsportfölj. MCLs prediktioner undersöks också för att se om de betalda och de kända skadekostnaderna konvergerar. En bootstrapmodell baserad på MCL (BMCL) undersöks för att utvärdera om möjligheterna att estimera täthetsfunktionen (probability density function, PDF) av framtida skadekostnader och av ”observable claim development results (OCDR)”. Resultaten visar att MCLs estimerade betalda och kända skadekostnader konvergerar. Resultaten visar även att när man evaluerar alla dataseten så ger MCL i genomsnitt bättre prediktioner än CLM med betald data, men ingen förbättring kan ses med CLM med känd skadekostnadsdata. Vidare visar resultaten även att genom att bara titta på dataset som uppfyller vissa krav, eller genom att bara använda olycksår efter 1999, så ökar andelen dataset där MCL ger bättre prediktioner än CLM.Vid evaluering av BMCL ser man att den kan producera estimerade PDF:er för ultimo-reserver och OCDR:er, men att medelvärdet av ultimo-reserv prediktionerna från BMCL inte konvergerar mot MCL-prediktionerna och att medelvärdet av OCDR:erna inte konvergerar mot noll. För att få rätt konvergens så centreras OCDR PDF:erna och ultimo-reservernas medelvärden sätts till motsvarande MCL-prediktionens värde genom multiplikation.

Skewness and kurtosis are measures used to describe shape characteristics of distributions. In this thesis, we examine the interval estimates about the skewness and kurtosis by using jackknife empirical likelihood (JEL), adjusted JEL, extended JEL, traditional bootstrap, percentile bootstrap, and BCa bootstrap methods. The limiting distribution of the JEL ratio is the standard chi-squared distribution. The simulation study of this thesis makes a comparison of different methods in terms of the coverage probabilities and interval lengths under the standard normal distribution and exponential distribution. The proposed adjusted JEL and extended JEL perform better than the other methods. Finally we illustrate the proposed JEL methods and different bootstrap methods with three real data sets.

Conformal Field Theories (CFT) are Quantum Field Theories characterized by enhanced (conformal) symmetries. They are interesting to Theoretical Physicists because they occur at critical points in phase transitions of various systems and also in the world sheet formulation of String Theory. CFTs allow Operator Product Expansion (OPE) in their correlators. The idea of Conformal Bootstrap is to solely use the conformal symmetries and crossing symmetry in the OPE to solve a conformal led theory and not explicitly use a lagrangian. Solving a CFT is equivalent to obtaining the anomalous dimensions and OPE coe client’s of the operators. The work presented in this thesis shows how ideas of bootstrap can be used to get analytic results for dimensions and OPE coe client’s of various operators in CFTs. In the conventional bootstrap program, the OPE in the direct (s-) channel is compared with the OPE of a crossed (t-) channel. This requirement of crossing symmetry is called the bootstrap equation. The flow of logic is somewhat reversed in the \new" idea that is formulated in this thesis. The trick is to expand a CFT correlator in terms of Witten diagrams, in all channels. This is a manifestly crossing symmetric description, and is in contrast to the usual expansion in terms of conformal blocks, which is in only one channel. For convenience we work with the Mellin transforms of Witten diagrams. For consistency of the Witten diagrams expansion with the conformal block expansion in a certain channel, we require the satisfaction of some equations, which we call the bootstrap equations in Mellin space. This scheme was rest chalked out by Polyakov in 1973, where he proposed the use of \unitary amplitudes" to expand a correlator. The unitary amplitudes had similar symmetry and analytic properties as the Witten diagrams. Even though he did not take his idea forward, replacing unitary amplitudes with Witten diagrams seems to work very well for obtaining analytic results. The working of bootstrap equations in Mellin space is demonstrated for the 4 Wilson-Fisher fixed point in d = 4 , O(N) theory at Wilson-Fisher point (in d = 4 ), as well as with large N (in general d), and large spin operators in strongly coupled and weakly coupled theories. For the case of global symmetry we have also analysed the somewhat unexplored case of cubic anisotropy. The results are obtained as perturbative series in , 1=N, or 1=` as applicable, and they are consistent with known results in literature. We also obtain various new results, for instance the OPE coe client’s of general higher spin operators. These results are otherwise very di cult to end from Feynman diagrams, but in this approach they come out very simply, essentially by solving some algebraic equations. We also show the use of the conventional bootstrap strategy, for analytically obtaining anomalous dimensions of large spin operators having higher twists, in a O(N) theory, by working in the light cone limit. One can question the validity of the proposal of using Witten diagrams to expand a correlator. One such issue is convergence of the sum over Witten diagrams. Convergence can be shown to hold for the operator spectrum we have worked with. Also there are operators that might upset convergence under some conditions. Resolutions of such cases, and ways to improve convergence have also been discussed. The conventional bootstrap method has been very successful in giving numerical results in nonpertur-bative CFTs, like the 3 dimensional Ising model. Numerical analysis can also be made possible with the new bootstrap in Mellin space approach. Having a convergent basis of expansion improves the prospect of numeric. The goal is to formulate a bootstrap scheme that, under a single framework, can make most of all the CFT properties. It should be systematic, so that one can obtain anomalous dimensions and OPE coe client’s of all operators up to any desired order, and works for all strongly/weakly coupled and perturbative/nonpertur-bative CFTS, both analytically and numerically. Finally, the use of Witten diagrams also indicates the possibility of Ising CFT or weakly coupled CFTs having connections with AdS/CFT, and hence String Theory. It does seem we have a right direction towards achieving our goal.

Dans le cadre de cette étude, il est question de prédire les primes de risque de marché pour les États-Unis et le Canada sur un horizon d’un mois en se basant sur les données économiques des 20 dernières années. En se basant sur les modèles élaborés précédemment dans la littérature, ce mémoire a pour but d’effectuer des prévisions plus précises que celles générées précédemment. Ainsi, on observe que l’ensemble des modèles retenus, tant univariés, multivariés ou par agrégation sectorielle obtiennent un pouvoir explicatif supérieur au modèle naïf, et ce tant pour le marché américain que le marché canadien. Pour le marché américain, le modèle multivarié Stepwise Backward obtient la meilleure performance du groupe étudié avec un R[indice supérieur 2] de 0.10714 dans un contexte In-Sample et un R[indice supérieur 2] de 0.22284 dans un contexte Out-Of-Sample selon le test de McCracken (2007). Ce modèle permet donc d’expliquer 22.28% de la variation mensuelle de la prime de risque américaine dans le contexte de l’échantillon observé dans cette étude. Le modèle en question est composé des variables économiques représentant les variations mensuelles au niveau de l’inflation, de la masse monétaire M2, ainsi que du dernier taux journalier du mois observé pour les obligations gouvernementales ayant une échéance de deux ans, cinq ans et dix ans. Pour le marché canadien, le modèle multivarié Stepwise Forward obtient la meilleure performance du groupe étudié dans un contexte In-Sample avec un R[indice supérieur 2] de 0.07760 selon le test de McCracken (2007). Cependant, dans un contexte Out-Of-Sample, le modèle de prévision par agrégation sectorielle élaboré à la section 4.4 obtient de loin la meilleure performance avec un R[indice supérieur 2] de 0.17773 selon le test de McCracken (2007), permettant donc d’expliquer 17.77% de la variation mensuelle de la prime de risque canadienne. La performance accrue des modèles de prévision dans un contexte Out-Of-Sample semble provenir d’une meilleure performance notamment dans les premières années d’observation, (2001 à 2007) grâce à l’exclusion des grandes variations affectant les dernières années de la période d’observation (2008 à 2011).