In this thesis we investigate some properties of market making and statistical arbitrage applied to High Frequency Trading (HFT). Using the Hamilton-Jacobi-Bellman(HJB) model developed by Guilbaud, Fabien and Pham, Huyen in 2012, we studied how market making works to obtain optimal strategy during limit order and market order. Also we develop the best investment strategy through Moving Average, Exponential Moving Average, Relative Strength Index, Sharpe Ratio.

In this thesis, we study a class of reaction-diffusion equations of the form $\frac{\partial u}{\partial t} = \mathcal{L}u + \phi u - \tfrac{1}{k} u^{k+1}$ where $\mathcal{L}$ is the stochastic generator of a Markov process, $\phi$ is a function of the space variables and $k\in \mathbb{R}\backslash\{0\}$. An important example, in the case when $k > 0$, is equations of the FKPP-type. We also give an example from the theory of utility maximisation problems when such equations arise and in this case $k < 0$. We introduce a new representation, for the solution of the equation, as the optimal value of an optimal control problem. We also give a second representation which can be seen as a dual problem to the first optimisation problem. We note that this is a new type of dual problem and we compare it to the standard Lagrangian dual formulation. By choosing controls in the optimisation problems we obtain upper and lower bounds on the solution to the PDE. We use these bounds to study the speed of the wave front of the PDE in the case when $\mathcal{L}$ is the generator of a suitable Lévy process.

Optimizing routes for multiple autonomous vehiclesin complex traffic situations can lead to improved efficiency intraffic. Attempting to solve these optimization problems centrally,i.e. for all vehicles involved, often lead to algorithms that exhibitthe curse of dimensionality: that is, the computation time andmemory needed scale exponentially with the number of vehiclesresulting in infeasible calculations for moderate number ofvehicles. However, using a numerical framework called tensordecomposition one can calculate and store solutions for theseproblems in a more manageable way. In this project, we investi-gate different tensor decomposition methods and correspondingalgorithms for solving optimal control problems, by evaluatingtheir accuracy for a known solution. We also formulate complextraffic situations as optimal control problems and solve them.We do this by using the best tensor decomposition and carefullyadjusting different cost parameters. From these results it canbe concluded that the Sequential Alternating Least Squaresalgorithm used with canonical tensor decomposition performedthe best. By asserting a smooth cost function one can solve certainscenarios and acquire satisfactory solutions, but it requiresextensive testing to achieve such results, since numerical errorsoften can occur as a result of an ill-formed problem. / Att optimera färdvägen för flertalet au-tonoma fordon i komplexa trafiksituationer kan leda till effekti-vare trafik. Om man försöker lösa dessa optimeringsproblemcentralt, för alla fordon samtidigt, leder det ofta till algorit-mer som uppvisar The curse of dimensionality, vilket är då beräkningstiden och minnes-användandet växer exponentielltmed antalet fordon. Detta gör många problem olösbara för endasten måttlig mängd fordon. Däremot kan sådana problem hanterasgenom numeriska verktyg så som tensornedbrytning. I det här projektet undersöker vi olika metoder för tensornedbrytningoch motsvarandes algoritmer för att lösa optimala styrproblem,genom att jämföra dessa för ett problem med en känd lösning.Dessutom formulerar vi komplexa trafiksituationer som optimalastyrproblem för att sedan lösa dem. Detta gör vi genom attanvända den bästa tensornedbrytningen och genom att noggrantanpassa kostnadsparametrar. Från dessa resultat framgår det att Sequential Alternating Least Squaresalgoritmen, tillsammans medkanonisk tensornedbrytning, överträffade de andra algoritmersom testades. De komplexa trafiksituationerna kan lösas genomatt ansätta släta kostnadsfunktioner, men det kräver omfattandetestning för att uppnå sådana resultat då numeriska fel lätt kan uppstå som ett resultat av dålig problemformulering. / Kandidatexjobb i elektroteknik 2020, KTH, Stockholm

Ce mémoire de maîtrise traite de la théorie de la ruine, et plus spécialement des modèles actuariels avec surplus dans lesquels sont versés des dividendes. Nous étudions en détail un modèle appelé modèle gamma-omega, qui permet de jouer sur les moments de paiement de dividendes ainsi que sur une ruine non-standard de la compagnie. Plusieurs extensions de la littérature sont faites, motivées par des considérations liées à la solvabilité. La première consiste à adapter des résultats d’un article de 2011 à un nouveau modèle modifié grâce à l’ajout d’une contrainte de solvabilité. La seconde, plus conséquente, consiste à démontrer l’optimalité d’une stratégie de barrière pour le paiement des dividendes dans le modèle gamma-omega. La troisième concerne l’adaptation d’un théorème de 2003 sur l’optimalité des barrières en cas de contrainte de solvabilité, qui n’était pas démontré dans le cas des dividendes périodiques. Nous donnons aussi les résultats analogues à l’article de 2011 en cas de barrière sous la contrainte de solvabilité. Enfin, la dernière concerne deux différentes approches à adopter en cas de passage sous le seuil de ruine. Une liquidation forcée du surplus est mise en place dans un premier cas, en parallèle d’une liquidation à la première opportunité en cas de mauvaises prévisions de dividendes. Un processus d’injection de capital est expérimenté dans le deuxième cas. Nous étudions l’impact de ces solutions sur le montant des dividendes espérés. Des illustrations numériques sont proposées pour chaque section, lorsque cela s’avère pertinent. / This master thesis is concerned with risk theory, and more specifically with actuarial surplus models with dividends. We focus on an important model, called the gamma-omega model, which is built to enable the study of both periodic dividend distributions and a non-standard type of ruin. We make several new extensions to this model, which are motivated by solvency considerations. The first one consists in adapting results from a 2011 paper to a new model built on the assumption of a solvency constraint. The second one, more elaborate, consists in proving the optimality of a barrier strategy to pay dividends in the gamma-omega model. The third one deals with the adaptation of a 2003 theorem on the optimality of barrier strategies in the case of solvency constraints, which was not proved right in the periodic dividend framework. We also give analogous results to the 2011 paper in case of an optimal barrier under the solvency constraint. Finally, the last one is concerned with two non-traditional ways of dealing with a ruin event. We first implement a forced liquidation of the surplus in parallel with a possibility of liquidation at first opportunity in case of bad prospects for the dividends. Secondly, we deal with injections of capital into the company reserve, and monitor their implications on the amount of expected dividends. Numerical illustrations are provided in each section, when relevant.

During the work of this research project we were interested in mathematical techniques that give us an insight to the following questions: How do we understand the trading decisions made by a manager of a hedge fund and what influences these decisions? In what way does an illiquid market affect these decisions and the performance of the fund? And how does the payment scheme affect the investor's decisions? Based on existing work on hedge fund management, we start with a fund that can be modelled with one risky investment and one riskless investment. Next, subject to the hedge fund special reward scheme we maximise the expected utility of wealth of the manager, by controlling the percentage invested in the risky investment, namely the portfolio. We use stochastic control techniques to derive a partial differential equation (PDE) and numerically obtain its corresponding viscosity solution, which provides a weak notion of solutions to these PDEs. This is then taken to a liquidity constrained scenario, to compare the behaviour of the two scenarios. Using the same approach as before we notice that due to the liquidity restriction we cannot use a simple model to combine the risky and riskless investments as a total amount, and hence the PDE is one order higher than before. We then model an investor who is investing in the hedge fund subject to the manager's optimal portfolio decisions, with similar mathematical tools as before. Comparisons between the investor's expected utility of wealth and the utility of having the money invested in the risk-free investment suggests that, in some cases, the investor is paying more to the manager than the return he is receiving for having invested in the hedge fund, compared to a risk-free investment. For that reason we propose a strategic game where the manager's action is to allocate the money between the two assets and the investor's action is to add money to the fund when he expects profit. The result is that the investor profits from the option to reinvest in the fund, although in some extreme cases the actions of the manager make the investor receive a negative value for having the option.

In the first part of our work we focus on the model of the optimal consumption with a random income. We provide the three dimensional equation for this model, demonstrate the reduction to the two dimensional case and provide for two different utility functions the full point-symmetries' analysis of the equations. We also demonstrate that for the logarithmic utility there exists a unique and smooth viscosity solution the existence of which as far as we know was never demonstrated before. In the second part of our work we develop the concept of the empirical liquidity measure. We provide the retrospective view of the works on this issue, discuss the proposed definitions and develop our own empirical measure based on the intuitive mathematical model and comprising several features of the definitions that existed before. Then we verify the measure provided on the real data from the market and demonstrate the advantages of the proposed value for measuring the illiquidity.

Guaranteed Minimum Withdrawal Benefits(GMWB) have become popular riders on variable annuities. The pricing of a GMWB contract was originally formulated as a singular stochastic control problem which results in a Hamilton Jacobi Bellman (HJB) Variational Inequality (VI). A penalty method method can then be used to solve the HJB VI. We present a rigorous proof of convergence of the penalty method to the viscosity solution of the HJB VI assuming the underlying asset follows a Geometric Brownian Motion. A direct control method is an alternative formulation for the HJB VI. We also extend the HJB VI to the case of where the underlying asset follows a Poisson jump diffusion. The HJB VI is normally solved numerically by an implicit method, which gives rise to highly nonlinear discretized algebraic equations. The classic policy iteration approach works well for the Geometric Brownian Motion case. However it is not efficient in some circumstances such as when the underlying asset follows a Poisson jump diffusion process. We develop a combined fixed point policy iteration scheme which significantly increases the efficiency of solving the discretized equations. Sufficient conditions to ensure the convergence of the combined fixed point policy iteration scheme are derived both for the penalty method and direct control method. The GMWB formulated as a singular control problem has a special structure which results in a block matrix fixed point policy iteration converging about one order of magnitude faster than a full matrix fixed point policy iteration. Sufficient conditions for convergence of the block matrix fixed point policy iteration are derived. Estimates for bounds on the penalty parameter (penalty method) and scaling parameter (direct control method) are obtained so that convergence of the iteration can be expected in the presence of round-off error.

Guaranteed Minimum Withdrawal Benefits(GMWB) have become popular riders on variable annuities. The pricing of a GMWB contract was originally formulated as a singular stochastic control problem which results in a Hamilton Jacobi Bellman (HJB) Variational Inequality (VI). A penalty method method can then be used to solve the HJB VI. We present a rigorous proof of convergence of the penalty method to the viscosity solution of the HJB VI assuming the underlying asset follows a Geometric Brownian Motion. A direct control method is an alternative formulation for the HJB VI. We also extend the HJB VI to the case of where the underlying asset follows a Poisson jump diffusion. The HJB VI is normally solved numerically by an implicit method, which gives rise to highly nonlinear discretized algebraic equations. The classic policy iteration approach works well for the Geometric Brownian Motion case. However it is not efficient in some circumstances such as when the underlying asset follows a Poisson jump diffusion process. We develop a combined fixed point policy iteration scheme which significantly increases the efficiency of solving the discretized equations. Sufficient conditions to ensure the convergence of the combined fixed point policy iteration scheme are derived both for the penalty method and direct control method. The GMWB formulated as a singular control problem has a special structure which results in a block matrix fixed point policy iteration converging about one order of magnitude faster than a full matrix fixed point policy iteration. Sufficient conditions for convergence of the block matrix fixed point policy iteration are derived. Estimates for bounds on the penalty parameter (penalty method) and scaling parameter (direct control method) are obtained so that convergence of the iteration can be expected in the presence of round-off error.

In this thesis, an optimal investment problem is studied for an investor who can only invest in a financial market modelled by an Itô-Lévy process; with one risk free (bond) and one risky (stock) investment possibility. We present the dynamic programming method and the associated Hamilton-Jacobi-Bellman (HJB) equation to explicitly solve this problem. It is shown that with purification and simplification to the standard jump diffusion process, closed form solutions for the optimal investment strategy and for the value function are attainable. It is also shown that, an explicit solution can be obtained via a finite training of a neural network using Stochastic gradient descent (SGD) for a specific case.

In this thesis, we develop partial differential equation (PDE) based numerical methods to solve certain optimal stochastic control problems in finance. The value of a stochastic control problem is normally identical to the viscosity solution of a Hamilton-Jacobi-Bellman (HJB) equation or an HJB variational inequality. The HJB equation corresponds to the case when the controls are bounded while the HJB variational inequality corresponds to the unbounded control case. As a result, the solution to the stochastic control problem can be computed by solving the corresponding HJB equation/variational inequality as long as the convergence to the viscosity solution is guaranteed. We develop a unified numerical scheme based on a semi-Lagrangian timestepping for solving both the bounded and unbounded stochastic control problems as well as the discrete cases where the controls are allowed only at discrete times. Our scheme has the following useful properties: it is unconditionally stable; it can be shown rigorously to converge to the viscosity solution; it can easily handle various stochastic models such as jump diffusion and regime-switching models; it avoids Policy type iterations at each mesh node at each timestep which is required by the standard implicit finite difference methods. In this thesis, we demonstrate the properties of our scheme by valuing natural gas storage facilities---a bounded stochastic control problem, and pricing variable annuities with guaranteed minimum withdrawal benefits (GMWBs)---an unbounded stochastic control problem. In particular, we use an impulse control formulation for the unbounded stochastic control problem and show that the impulse control formulation is more general than the singular control formulation previously used to price GMWB contracts.