<p>We consider one transaction costs model which was suggested by Cetin, Jarrow and Protter (2004) for an illiquid market. In this case the hedging strategy of programming traders can affect the assets prises. We study the corresponding partial differential equation (PDE) which is a non-linear Black-Scholes equation for illiquid markets. We use the Lie group analysis to determine the symmetry group of this equations. We present the Lie algebra of the Lie point transformations, the complete symmetry group and invariants. For different subgroups of the main symmetry group we describe the corresponding invariants. We use these invariants to reduce the PDE under investigation to ordinary differential equations (ODE). Solutions of these ODE's are subgroup-invariant solutions of the non-linear Black-Scholes equation. For some classes of those ODE's we find exact solutions and studied their properties.</p><p>% reduce non-linear PDE to ODE's. To some ODE's we find exact solutions.</p><p>%In this work we are studying one model for pricing derivatives in illiquid market. We discuss it structure and properties. Make a symmetry reduction for the PDE corresponding our model.</p>

We consider one transaction costs model which was suggested by Cetin, Jarrow and Protter (2004) for an illiquid market. In this case the hedging strategy of programming traders can affect the assets prises. We study the corresponding partial differential equation (PDE) which is a non-linear Black-Scholes equation for illiquid markets. We use the Lie group analysis to determine the symmetry group of this equations. We present the Lie algebra of the Lie point transformations, the complete symmetry group and invariants. For different subgroups of the main symmetry group we describe the corresponding invariants. We use these invariants to reduce the PDE under investigation to ordinary differential equations (ODE). Solutions of these ODE's are subgroup-invariant solutions of the non-linear Black-Scholes equation. For some classes of those ODE's we find exact solutions and studied their properties. % reduce non-linear PDE to ODE's. To some ODE's we find exact solutions. %In this work we are studying one model for pricing derivatives in illiquid market. We discuss it structure and properties. Make a symmetry reduction for the PDE corresponding our model.

In dieser Dissertation lösen wir eine Klasse stochastischer Kontrollprobleme und konstruieren optimale Handelsstrategien in illiquiden Märkten. In Kapitel 1 betrachten wir einen Investor, der sein Portfolio nahe an einer stochastischen Zielfunktion halten möchte. Gesucht ist eine Strategie (aus aktiven und passiven Orders), die die Abweichung vom Zielportfolio und die Handelskosten minimiert. Wir zeigen Existenz und Eindeutigkeit einer optimalen Strategie. Wir beweisen eine Version des stochastischen Maximumprinzips und leiten damit ein Kriterium für Optimalität mittels einer gekoppelten FBSDE her. Wir beweisen eine zweite Charakterisierung mittels Kauf- und Verkaufregionen. Das Portfolioliquidierungsproblem wird explizit gelöst. In Kapitel 2 verallgemeinern wir die Klasse der zulässigen Strategien auf singuläre Marktorders. Wie zuvor zeigen wir Existenz und Eindeutigkeit einer optimalen Strategie. Im zweiten Schritt beweisen wir eine Version des Maximumprinzips im singulären Fall, die eine notwendige und hinreichende Optimalitätsbedingung liefert. Daraus leiten wir eine weitere Charakterisierung mittels Kauf-, Verkaufs- und Nichthandelsregionen ab. Wir zeigen, dass Marktorders nur benutzt werden, wenn der Spread klein genug ist. Wir schließen dieses Kapitel mit einer Fallstudie über Portfolioliquidierung ab. Das dritte Kapitel thematisiert Marktmanipulation in illiquiden Märkten. Wenn Transaktionen einen Einfluß auf den Aktienpreis haben, dann können Optionsbesitzer damit den Wert ihres Portfolios beeinflussen. Wir betrachten mehrere Agenten, die europäische Derivate halten und den Preis des zugrundeliegenden Wertpapiers beeinflussen. Wir beschränken uns auf risikoneutrale und CARA-Investoren und zeigen die Existenz eines eindeutigen Gleichgewichts, das wir mittels eines gekoppelten Systems nichtlinearer PDEs charakterisieren. Abschließend geben wir Bedingungen an, wie diese Art von Marktmanipulation verhindert werden kann. / In this thesis we study a class of stochastic control problems and analyse optimal trading strategies in limit order markets. The first chapter addresses the problem of curve following. We consider an investor who wants to keep his stock holdings close to a stochastic target function. We construct the optimal strategy (comprising market and passive orders) which balances the penalty for deviating and the cost of trading. We first prove existence and uniqueness of an optimal control. The optimal trading strategy is then characterised in terms of the solution to a coupled FBSDE involving jumps via a stochastic maximum principle. We give a second characterisation in terms of buy and sell regions. The application of portfolio liquidation is studied in detail. In the second chapter, we extend our results to singular market orders using techniques of singular stochastic control. We first show existence and uniqueness of an optimal control. We then derive a version of the stochastic maximum principle which yields a characterisation of the optimal trading strategy in terms of a nonstandard coupled FBSDE. We show that the optimal control can be characterised via buy, sell and no-trade regions. We describe precisely when it is optimal to cross the bid ask spread. We also show that the controlled system can be described in terms of a reflected BSDE. As an application, we solve the portfolio liquidation problem with passive orders. When markets are illiquid, option holders may have an incentive to increase their portfolio value by using their impact on the dynamics of the underlying. In Chapter 3, we consider a model with competing players that hold European options and whose trading has an impact on the price of the underlying. We establish existence and uniqueness of equilibrium results and show that the equilibrium dynamics can be characterised in terms of a coupled system of non-linear PDEs. Finally, we show how market manipulation can be reduced.

<p>This master project is dedicated to the analysis of one of the nancialmarket models in an illiquid market. This is a nonlinear model. Using analytical methods we studied the symmetry properties of theequation which described the given model. We called this equation aSchonbucher-Wilmott equation or the main equation. We have foundinnitesimal generators of the Lie algebra, containing the informationabout the symmetry group admitted by the main equation. We foundthat there could be dierent types of the unknown function g, whichwas located in the main equation, in particular four types which admitsricher symmetry group. According to the type of the function gthe equation was split up into four PDEs with the dierent Lie algebrasin each case. Using the generators we studied the structure ofthe Lie algebras and found optimal systems of subalgebras. Then weused the optimal systems for dierent reductions of the PDE equationsto some ODEs. Obtained ODEs were easier to solve than the correspondingPDE. Thereafter we proceeded to the solution of the desiredSchonbucher-Wilmott equation. In the project we were guided by thepapers of Bank, Baum [1] and Schonbucher, Wilmott [2]. In these twopapers authors introduced distinct approaches of the analysis of thenonlinear model - stochastic and dierential ones. Both approaches leadunder some additional assumptions to the same nonlinear equation - the main equation.</p>

This master project is dedicated to the analysis of one of the nancialmarket models in an illiquid market. This is a nonlinear model. Using analytical methods we studied the symmetry properties of theequation which described the given model. We called this equation aSchonbucher-Wilmott equation or the main equation. We have foundinnitesimal generators of the Lie algebra, containing the informationabout the symmetry group admitted by the main equation. We foundthat there could be dierent types of the unknown function g, whichwas located in the main equation, in particular four types which admitsricher symmetry group. According to the type of the function gthe equation was split up into four PDEs with the dierent Lie algebrasin each case. Using the generators we studied the structure ofthe Lie algebras and found optimal systems of subalgebras. Then weused the optimal systems for dierent reductions of the PDE equationsto some ODEs. Obtained ODEs were easier to solve than the correspondingPDE. Thereafter we proceeded to the solution of the desiredSchonbucher-Wilmott equation. In the project we were guided by thepapers of Bank, Baum [1] and Schonbucher, Wilmott [2]. In these twopapers authors introduced distinct approaches of the analysis of thenonlinear model - stochastic and dierential ones. Both approaches leadunder some additional assumptions to the same nonlinear equation - the main equation.