There is a little doubt that, for a decade, risk measurement has become one of the most important topics in finance. Indeed, it is natural to observe such a development, since in the last ten years, huge amounts of financial transactions ended with severe losses due to severe convulsions in financial markets. Value at risk, as the most widely used risk measure, fails to quantify the risk of a position accurately in many situations. For this reason a number of consistent risk measures have been introduced in the literature. The main aim of this study is to present and compare coherent, convex, conditional convex and some other risk measures both in theoretical and practical settings.

One of the financial risks an agent has to deal with is market risk. Market risk is caused by the uncertainty attached to asset values. There exit various measures trying to model market risk. The most widely accepted one is Value-at- Risk. However Value-at-Risk does not encourage portfolio diversification in general, whereas a consistent risk measure has to do so. In this work, risk measures satisfying these consistency conditions are examined within theoretical basis. Different types of coherent and convex risk measures are investigated. Moreover the extension of coherent risk measures to multiperiod settings is discussed.

Mit der vorliegenden Dissertation wollen wir einen Beitrag zur Theorie der konvexen Risikomaße und ihrer Dynamik leisten. Im Kapitel 1 beschäftigen wir uns zunächst mit unbedingten konvexen Risikomaßen. Wir erläutern die Eigenschaften dieser Funktionale und geben einen Überblick über die Möglichkeiten ihrer Darstellung. Anschließend diskutieren wir das Problem ihrer Fortsetzbarkeit. Im Kapitel 2 erklären wir, wie sich die Darstellungssätze auf bedingte konvexe Risikomaße übertragen lassen, und untersuchen, unter welchen Voraussetzungen eine reguläre bedingte Darstellung existiert. Auf polnischen Räumen beweisen wir die Existenz auf den Klassen der halbstetigen Funktionen. Für das bedingte Average Value at Risk zeigen wir, dass durch eine zugehörige Familie von unbedingten AVaR-Risikomaßen eine reguläre bedingte Darstellung sogar auf der Klasse aller beschränkten Auszahlungsprofile gegeben ist. Im Kapitel 3 untersuchen wir die intertemporale Struktur von dynamischen konvexen Risikomaßen. Zunächst analysieren wir verschiedene Formen der Akzeptanz- und Ablehnungskonsistenz, welche einem zeitlich rückwärts gerichteten Ansatz der Risikobewertung entsprechen und in der Regel zur Konstruktion von dynamischen konvexen Risikomaßen zu einer vorgegebenen Filtration verwendet werden. Als Alternative formulieren wir einen vorwärts gerichteten Ansatz, bei dem jedes bedingte konvexe Risikomaß als eine Konsequenz aus der vorherigen Risikobewertung und der eingehenden Information konstruiert wird. Dann diskutieren wir Aktualisierungsvorschriften für konvexe Risikomaße. Wir überprüfen, inwieweit die vorgestellten Bedingungen der zeitlichen Konsistenz in ihrer starken und schwachen Form oder die Bedingung der Konsequenz als Aktualisierungskriterium geeignet sind. In diesem Zusammenhang diskutieren wir abschließend auch das Problem der Unsicherheitsreduzierung nach dem Erhalt von Zusatzinformation. / This thesis is a contribution to the theory convex risk measures and their dynamics. In chapter 1 we consider unconditional convex risk measures. At first, we explain the properties of these functionals and present different possibilities of their representation. Then we discuss the extension problem for convex risk measures. In chapter 2 we study conditional convex risk measures and their representations. We also analyze under which conditions these functionals admit a regular conditional representation. On polish spaces we prove existence on the classes of semicontinuous functions. For the conditional Average Value at Risk, we show that a regular conditional representation is given by a corresponding family of unconditional AVaR risk measures on the class of all bounded payoff functions. In Chapter 3 we investigate the intertemporal structure of dynamic convex risk measures. We begin by considering different conditions of acceptance and rejection consistency which correspond to a backward approach of dynamic risk evaluation and which are used for the construction of dynamic convex risk measures with respect to some given filtration. We also introduce an alternative forward approach where each conditional convex risk measure is constructed as a consequence of the initial risk evaluation and the incoming information. Then we discuss update rules for convex risk measures. We analyze whether the conditions of strong and weak consistency and the condition of consecutivity are appropriate update criteria. In this context, we finally discuss how uncertainty may be reduced after receiving some additional information.

In dieser Arbeit werden verschiedene Eigenschaften von dynamischen konvexen Risikomaßen für beschränkte Zufallsvariablen untersucht. Dabei gehen wir vor allem der Frage nach, wie die Risikobewertungen in verschiedenen Zeitpunkten von einander abhängen, und wie sich solche Zeitkonsistenzeigenschaften in der Dynamik der Penalty-Funktionen und Risikoprozesse widerspiegeln. Im Kapitel 2 widmen wir uns zunächst der starken Zeitkonsistenz und charakterisieren diese mithilfe von Akzeptanzmengen, Penalty-Funktionen und einer gemeinsamen Supermartingaleigenschaft des Risikoprozesses und seiner Penalty-Funktion. Die Charakterisierung durch Penalty-Funktionen liefert eine explizite Form der Doob- und der Riesz-Zerlegung des Prozesses der Penalty-Funktionen. Anschließend führen wir einen schwächeren Begriff der Zeitkonsistenz ein, den wir Besonnenheit nennen. In Analogie zu dem zeitkonsistenten Fall charakterisieren wir Besonnenheit durch Akzeptanzmengen, Penalty-Funktionen und eine bestimmte Supermartingaleigenschaft. Diese Supermartingaleigenschaft gilt allgemeiner für alle beschränkten adaptierten Prozesse, die sich ohne zusätzliches Risiko aufrechterhalten lassen. Wir nennen solche Prozesse nachhaltig und beschreiben Nachhaltigkeit durch eine gemeinsame Supermartingaleigenschaft des Prozesses und der schrittweisen Penalty-Funktionen. Dieses Resultat kann als eine verallgemeinerte optionale Zerlegung unter konvexen Restriktionen gesehen werden. Mithilfe der Supermartingaleigenschaft identifizieren wir das stark zeitkonsistente dynamische Risikomaß, das aus jedem beliebigen Risikomaß rekursiv konstruiert werden kann, als den kleinsten Prozeß, der nachhaltig ist und den Endverlust minimiert. Diese Beschreibung liefert ein neues Argument für den Einsatz von zeitkonsistenten Risikomaßen. Im Kapitel 3 diskutieren wir das asymptotische Verhalten von zeitkonsistenten und von besonnenen Risikomaßen hinsichtlich der asymptotischen Sicherheit und der asymptotischen Präzision. Im Kapitel 4 werden die allgemeinen Ergebnisse aus den Kapiteln 2 und 3 anhand des entropischen Risikomaßes und des Superhedging-Preisprozesses veranschaulicht. / In this thesis we study various properties of a dynamic convex risk measure for bounded random variables. The main subject is to investigate possible interdependence of conditional risk assessments at different times and the manifestation of these time consistency properties in the dynamics of corresponding penalty functions and risk processes. In Chapter 2 we focus first on the strong notion of time consistency and characterize it in terms of penalty functions, acceptance sets and a joint supermartingale property of the risk measure and its penalty function. The characterization in terms of penalty functions provides the explicit form of the Doob and of the Riesz decomposition of the penalty function process for a time consistent risk measure. Then we introduce and study a weaker notion of time consistency, that we call prudence. Similar to the time consistent case, we characterize prudent dynamic risk measures in terms of acceptance sets, of penalty functions and by a certain supermartingale property. This supermartingale property holds more generally for any bounded adapted process that can be upheld without any additional risk. We call such processes sustainable, and we give an equivalent characterization of sustainability in terms of a combined supermartingale property of a process and one-step penalty functions. This result can be viewed as a generalized optimal decomposition under convex constraints. The supermartingale property allows us to characterize the strongly time consistent risk measure arising from any dynamic risk measure via recursive construction as the smallest process that is sustainable and covers the final loss. Thus our discussion provides a new reason for using strongly time consistent risk measures. In Chapter 3 we discuss the limit behavior of time consistent and of prudent risk measures in terms of asymptotic safety and of asymptotic precision. In the final Chapter 4 we illustrate the general results of Chapter 2 and Chapter 3 by examples. In particular we study the entropic dynamic risk measure and the superhedging price process under convex constraints.

The aim of this thesis is to present new results concerning duality in scalar optimization. We show how the theory can be applied to optimization problems arising in the theory of risk measures, portfolio optimization and machine learning. First we give some notations and preliminaries we need within the thesis. After that we recall how the well-known Lagrange dual problem can be derived by using the general perturbation theory and give some generalized interior point regularity conditions used in the literature. Using these facts we consider some special scalar optimization problems having a composed objective function and geometric (and cone) constraints. We derive their duals, give strong duality results and optimality condition using some regularity conditions. Thus we complete and/or extend some results in the literature especially by using the mentioned regularity conditions, which are weaker than the classical ones. We further consider a scalar optimization problem having single chance constraints and a convex objective function. We also derive its dual, give a strong duality result and further consider a special case of this problem. Thus we show how the conjugate duality theory can be used for stochastic programming problems and extend some results given in the literature. In the third chapter of this thesis we consider convex risk and deviation measures. We present some more general measures than the ones given in the literature and derive formulas for their conjugate functions. Using these we calculate some dual representation formulas for the risk and deviation measures and correct some formulas in the literature. Finally we proof some subdifferential formulas for measures and risk functions by using the facts above. The generalized deviation measures we introduced in the previous chapter can be used to formulate some portfolio optimization problems we consider in the fourth chapter. Their duals, strong duality results and optimality conditions are derived by using the general theory and the conjugate functions, respectively, given in the second and third chapter. Analogous calculations are done for a portfolio optimization problem having single chance constraints using the general theory given in the second chapter. Thus we give an application of the duality theory in the well-developed field of portfolio optimization. We close this thesis by considering a general Support Vector Machines problem and derive its dual using the conjugate duality theory. We give a strong duality result and necessary as well as sufficient optimality conditions. By considering different cost functions we get problems for Support Vector Regression and Support Vector Classification. We extend the results given in the literature by dropping the assumption of invertibility of the kernel matrix. We use a cost function that generalizes the well-known Vapnik's ε-insensitive loss and consider the optimization problems that arise by using this. We show how the general theory can be applied for a real data set, especially we predict the concrete compressive strength by using a special Support Vector Regression problem.

The aim of this thesis is to present new results concerning duality in scalar optimization. We show how the theory can be applied to optimization problems arising in the theory of risk measures, portfolio optimization and machine learning. First we give some notations and preliminaries we need within the thesis. After that we recall how the well-known Lagrange dual problem can be derived by using the general perturbation theory and give some generalized interior point regularity conditions used in the literature. Using these facts we consider some special scalar optimization problems having a composed objective function and geometric (and cone) constraints. We derive their duals, give strong duality results and optimality condition using some regularity conditions. Thus we complete and/or extend some results in the literature especially by using the mentioned regularity conditions, which are weaker than the classical ones. We further consider a scalar optimization problem having single chance constraints and a convex objective function. We also derive its dual, give a strong duality result and further consider a special case of this problem. Thus we show how the conjugate duality theory can be used for stochastic programming problems and extend some results given in the literature. In the third chapter of this thesis we consider convex risk and deviation measures. We present some more general measures than the ones given in the literature and derive formulas for their conjugate functions. Using these we calculate some dual representation formulas for the risk and deviation measures and correct some formulas in the literature. Finally we proof some subdifferential formulas for measures and risk functions by using the facts above. The generalized deviation measures we introduced in the previous chapter can be used to formulate some portfolio optimization problems we consider in the fourth chapter. Their duals, strong duality results and optimality conditions are derived by using the general theory and the conjugate functions, respectively, given in the second and third chapter. Analogous calculations are done for a portfolio optimization problem having single chance constraints using the general theory given in the second chapter. Thus we give an application of the duality theory in the well-developed field of portfolio optimization. We close this thesis by considering a general Support Vector Machines problem and derive its dual using the conjugate duality theory. We give a strong duality result and necessary as well as sufficient optimality conditions. By considering different cost functions we get problems for Support Vector Regression and Support Vector Classification. We extend the results given in the literature by dropping the assumption of invertibility of the kernel matrix. We use a cost function that generalizes the well-known Vapnik's ε-insensitive loss and consider the optimization problems that arise by using this. We show how the general theory can be applied for a real data set, especially we predict the concrete compressive strength by using a special Support Vector Regression problem.