Variance reduced Value at Risk Monte-Carlo simulations
|
|
- Magnus Warren
- 6 years ago
- Views:
Transcription
1 Variance reduced Value at Risk Monte-Carlo simulations Armin Müller 1 Discussion paper No. 500 November 24, 2016 Diskussionsbeiträge der Fakultät für Wirtschaftswissenschaft der FernUniversität in Hagen Herausgegeben vom Dekan der Fakultät Alle Rechte liegen bei dem Verfasser Abstract: Monte-Carlo simulations of the risk measure Value at Risk inherently involve standard errors that depend on the sample size N. In this article, we present a variance reduction technique for the estimation of loss probabilities using importance sampling. For a given sample size N, the method reduces the empirical variance of these loss probabilities by more than two orders of magnitude. Thus, it yields more accurate estimators than a standard Monte-Carlo simulation. 1 FernUniversität in Hagen, Lehrstuhl für angewandte Statistik und Methoden der empirischen Sozialforschung, D Hagen, Germany, armin.mueller@fernuni-hagen.de
2 1 Introduction Value at Risk (VaR) intends to represent the total risk of financial instruments. Monte-Carlo simulations are one method to compute Value at Risk estimators. These estimators naturally involve standard errors. These standard errors typically decrease with an increasing sample size N. Importance sampling is a variance reduction technique that is aimed at reducing the empirical variance of Monte-Carlo estimators for a given sample size. In this article we will apply importance sampling to the estimation of loss probabilities that can be interpreted as quantiles of loss distributions, i.e. as Value at Risk estimates. Our method reduces the empirical variance of these estimators significantly. Therefore, it yields more accurate estimators with lower standard errors than a standard Monte-Carlo simulation for a given sample size N. The technique employed has been introduced by Singer and has been applied to several other applications already, e.g. to option pricing or the Ginzburg-Landau model of superconductivity [1, 2]. Alternative methods for variance reduced VaR Monte-Carlo simulations already exist [3, 4]. The article is organized as follows: First, we introduce the concept of VaR in section 2. Section 3 introduces the basic concepts of importance sampling, derives a variance reduced VaR Monte-Carlo estimator and introduces suitable elasticity approximations required for the variance reduced Monte-Carlo simulation. In section 4, we present a numerical example that demonstrates the functionality of the presented approach. We discuss results in section 5. The article concludes in section 6. 2 The risk measure VaR VaR is a risk measure that aims to aggregate the entire risk of a portfolio with value X(t) into one single indicator. It coincides with the loss a portfolio will not exceed at a given confidence level α over a given period of time. Consequently, the VaR corresponds to the (1-α)-quantile of the loss distribution of the portfolio value [5]. Usually, the VaR measure is used to indicate the risk of an entire portfolio. However, in some cases VaR figures are calculated for single financial instruments. For example, regulatory authorities require financial institutions to provide individual VaR estimates for PRIIPs (Packaged Retail and Insurance-based Investment Products) 2. Going forward, we will focus on portfolios. Of course, this involves the 2 For details, see [6]. Descriptions of typical PRIIPs can be found at [7, 8]. 2
3 special case of portfolios that consist of one single financial instrument only. Several methods exist to provide VaR estimates [5]. The most popular approach is the non-parametric historical simulation. Historical data is employed to estimate the future value development of the considered portfolio. Also parametric approaches exist, e.g., the variance-covariance method. The focus of this article is a third also parametric approach, the Monte-Carlo simulation. This approach yields VaR estimators involving standard errors. The size of these standard errors is proportional to 1/ N where N is the sample size. 3 Importance sampling of VaR estimators Importance sampling is a powerful approach to reduce the standard error of Monte- Carlo estimators. In this section, we first briefly introduce the general concept of importance sampling. Subsequently, we derive a variance reduced VaR Monte-Carlo estimator and provide suitable elasticity approximations required for the presented importance sampling approach. 3.1 Basics of importance sampling The expectation value of a function h : R d R, X h(x) of a random variable X with probability density p is defined as α = IE p [h (X)] = h (x) p (x) dx. (1) An unbiased Monte-Carlo estimator of this quantity is ˆα p = 1 n N h (X i ) (2) i=1 for i.i.d. realizations X 1,..., X N of X. Expanding the integrand of equation (1) with any other probability density p leads to α = It can be shown, that the empirical variance of the unbiased Monte-Carlo estimator h (x) p (x) dx = h (x) p (x) [ p (x) p (x) dx = IE p h (X) p (X) ]. (3) p (X) ˆα p = 1 n N i=1 h (X i ) p (X i) p (X i ) (4) 3
4 vanishes, if, for a non-negative function h, we choose p (x) h(x) p(x). (5) By normalizing, the product h(x) p(x) can be transformed into a probability density [9]. An importance sampling approach introduced by Singer [1, 2] aims to conduct variance reduced Monte-Carlo simulations to calculate integrals of the form P (X(t), t) = h [X(T )] p[x(t ), T X(t), t]dx(t ) (6) where the random process X(t) follows an Itō differential equation of the following type: dx(t) = f[x(t)]dt + g[x(t)]dw (t) (7) In this equation, f is the drift vector, t the time, g the diffusion coefficient and W (t) a multivariate Wiener process. In a lengthy calculation, it can be shown that variance reduction can be achieved by adding an additional drift term to equation (7): dx(t) = (f[x(t)] + δf) dt + g[x(t)]dw (t) (8) For the optimal additional drift term δf, one obtains δf = Ω ln P (9) with the diffusion matrix Ω = gg T. The task of the additional drift term is to drag the stochastic process X(t) to areas where the density p described in equation (5) takes on high values [1, 2]. To conduct a Monte-Carlo simulation with the purpose of estimating the quantity P, a discretization of equation (8) using a Euler-Maruyama approximation [10] is required: X k+1 = X k + (f k + δf k ) τ + g k W k (10) The following estimator for the integral in equation (6) with sample size N and n discretization steps results: ˆP = 1 N { N n 1 ( ) } 1 exp 2 δf k T Ω k 1 δf k τ + δfk T Ω k 1 g k W k h[x in] i=1 k=0 (11) 4
5 The exponential function corresponds to the term p/p in equation (4). A detailed derivation of these formulae can be found in [2] and [11]. 3.2 A variance reduced VaR Monte-Carlo estimator The next step is to derive a variance reduced VaR Monte-Carlo estimator. We assume that the considered portfolio at time t has the value X(t). The total loss L of the portfolio s value in the interval [0, T ] is defined as L := X = [X(T ) X(0)]. (12) Our aim is to estimate the probability P, that in this interval the loss L exceeds a given level b, i.e. to estimate P (L > b). Using the indicator function I and the transition density 3 p[x(t ) X(0)], this quantity can be expressed as P [L > b X(0)] = 0 I(L > b) p[x(t ) X(0)] dx(t ). (13) For simplicity, we assume that the stochastic process X(t) follows a Geometric Brownian motion with drift rate f = rx and diffusion rate g = σx. This justifies 0 as the lower integration limit instead of. We also assume that X(t) follows a univariate stochastic process. However, also more complex, multivariate models, e.g., models with stochastic volatility, are compatible with the presented importance sampling approach 4. Consequently, using equations (9) and (10), we obtain the discretized stochastic differential equation X k+1 = X k + ( r + σ 2 ɛ k ) Xk τ + σx k W k (14) with the elasticity ɛ k = X k P (L > b X k ). (15) P (L > b X k ) X k Using equations (14) and (15) to simulate the portfolio value s trajectories, the following formula results from equation (11) yielding a variance reduced Monte- Carlo estimator of the probability that the total loss L exceeds the given level b: 3 Note that the transition probability also explicitly depends on the time. Therefore, we should write p = p[x(t ), T X(t), t]. However, for the purpose of compactness, we suppress the time argument. 4 For details, see [11, 12]. 5
6 ˆP (L > b X 0 ) = 1 N { N n 1 ( ) } 1 exp 2 σ2 ɛ 2 ik τ + σɛ ik W k I(L i > b) i=1 k=0 (16) 3.3 Elasticity approximations for variance reduced VaR estimates The elasticity ɛ k is unknown ex anteriori, as the probability P (L > b X k ) is generally an unknown quantity. Therefore, we conduct approximations of the elasticity. The quality of the achieved variance reduction highly depends on the specific approximation employed 5. This article focuses on the approximation by constant values and on the approximation by numerical integration 6. Approximation by constant values The least complex approach is to approximate the elasticity ɛ by a constant value. Equation (15) indicates that ɛ should take on only negative values as obviously P (L>b X k) X k < 0. (The higher the portfolio value, the lower the probability that the loss will exceed a given threshold.) Increasing the absolute value of the constant elasticity approximation has two counteracting effects: On the one hand, more trajectories of X(t) will result in a loss L exceeding b and therefore contribute to the Monte-Carlo estimator. On the other hand, the higher the absolute value of ɛ, the higher the discretization bias resulting from the discretized stochastic differential equation (14) will be. Approximation by numerical integration A more sophisticated approach is the approximation of ɛ by numerical integration. For each trajectory i and each discretization step k we calculate P (L > b X k ) and its partial derivative with respect to X k by numerical integration employing Gauss-Legendre quadrature. Based on equation (13) and using discrete time, with definition (12) the loss prob- 5 An analysis of several possible approximations can be found in [11]. 6 It is important to note, that equation (16) yields an unbiased estimator, independently of the specific choice of ɛ. The choice only influences the estimator s standard error, not its mean value. Of course, for very high absolute ɛ values, a discretization bias is introduced. 6
7 ability can be recast as follows: P (L > b X k ) = = 0 X0 b = 1 I(L > b) p[x n X k ] dx n 0 p[x n X k ] dx n X 0 b p[x n X k ] dx n (17) We assume that p[x n X k ] is the density of the log-normal distribution [5]. As mentioned above, in this article we only consider Geometric Brownian motions. In this special case, the assumption is accurate. However, also for more complex models of the stochastic process X(t) for the purpose of calculating ɛ this approach is permissible (compare footnote 6). Thus, we obtain p[x n X k ] = 1 1 2πσ2 (n k) t X n [ ln X n ln X k exp ( r σ2 2 2σ 2 (n k) t ) (n k) t ] 2. (18) The affine transformation [13] X n = X 0 b y 1 y X 0 b f(x n )dx n = 1 1 ( 2 (1 y) f X 2 0 b y ) dy 1 y (19) yields P (L > b X k ) = 1 exp ( ( ln 1 1 2πσ2 (n k) t 1 ) X 0 b + 1+y 1 y 2 1 (1 y) 2 X 0 b + 1+y 1 y ( ) ) 2 ln X k r σ2 (n k) t 2 2σ 2 (n k) t dy. (20) For the purpose of compactness, we abbreviate ( ξ := ln X 0 b y ) ln X k 1 y ) (r σ2 (n k) t (21) 2 7
8 and obtain P (L > b X k ) = 1 2πσ2 (n k) t 1 (1 y) 2 X 0 b + 1+y 1 y { } ξ 2 exp dy. 2σ 2 (n k) t (22) This term can be evaluated by means of Gauss-Legendre quadrature [14, 15]. Abbreviating ( ξ i := ln X 0 b x ) i ln X k 1 x i with n GL abscissae x i and weights w i, we obtain ) (r σ2 (n k) t (23) 2 P (L > b X k ) 1 1 2πσ2 (n k) t { exp ξ 2 i n GL i=1 2σ 2 (n k) t 2 1 (1 x i ) 2 X 0 b + 1+x i 1 x } i w i. (24) Analogously, for the partial derivative, by applying Leibniz rule for parameter integrals [16] the following expression results: P (L > b X k ) X k = 2πσ2 (n k) t 1 (1 y) 2 X 0 b + 1+y 1 y { } ξ 2 exp dy X k 2σ 2 (n k) t = 2πσ2 (n k) t 1 (1 y) 2 X 0 b + 1+y 1 y { } ξ 2 ξ 1 exp dy 2σ 2 (n k) t σ 2 (n k) t X k (25) Note that for k = 0 additional terms appear, as ( ) ln X 0 b + 1+y 1 y X 0 and ( X 0 b + 1+y 1 y X 0 do not vanish. For reasons of simplicity, we suppress these terms and use the same expression for the first discretization step as for all the other discretization steps. ) 1 8
9 Again, we conduct a Gauss-Legendre quadrature and obtain P (L > b X k ) X k n GL πσ2 (n k) t (1 x i=1 i ) 2 X 0 b + 1+x i 1 x { } i ξi 2 ξ i 1 exp w 2σ 2 (n k) t σ 2 i. (n k) t X k (26) With these quadrature results we can calculate ɛ ik for each trajectory i and each discretization step k from equation (15). Before presenting numerical results, two comments should be made: The calculation of N n different ɛ ik values involving the Gauss-Legendre quadrature of at least two integrals each is very demanding with regards to computational power. An approach that can significantly reduce the required amount of arithmetical operations is to first calculate an ɛ-grid as a function of the current portfolio value X k and the remaining discretized time (n k) t. The required ɛ ik for each trajectory i and each discretization step k can then be obtained by linear intrapolation or by cubic spline intrapolation from this grid. In the Black-Scholes model involving a Geometric Brownian motion for the portfolio value X(t), Monte-Carlo simulations are of course not required to calculate quantiles of the log-normal distribution. However, the importance sampling approach presented in this article also works well for more complex models, e.g., models with stochastic volatility. In this case, for the calculation of ɛ, the assumption of a log-normal transition density is a well-working approximation 7. 4 Numerical results We now present a numerical example that demonstrates the ability of the introduced approach to significantly reduce the empirical variance of VaR Monte-Carlo estimators. We consider the following setting: The considered portfolio X(t) has an initial value of X(0) = 50. The portfolio s value follows a Geometric Brownian motion with interest rate r = 0.05 and volatility σ = Compare [11], section 3.7 for details. 9
10 b = 10 Benchmark Constant approximation Integration ɛ = 5 ɛ = 7.5 ɛ = 15 ɛ = 20 ˆP (L > b) 9.30% 10.36% 10.79% 10.77% 13.02% 10.33% ˆσ ˆP 0.92% 0.45% 0.42% 0.85% 3.35% 0.13% ˆσ 2ˆP 8.5E E E E E E-06 VarRatio Table 1: Variance reduced Monte-Carlo simulation of the probability that the portfolio loss exceeds b = 10. b = 20 Benchmark Constant approximation Integration ɛ = 5 ɛ = 7.5 ɛ = 15 ɛ = 20 ˆP (L > b) 0.30% 0.37% 0.37% 0.31% 0.38% 0.34% ˆσ ˆP 0.17% 0.05% 0.03% 0.02% 0.03% 0.01% ˆσ 2ˆP 3.0E E E E E E-09 VarRatio Table 2: Variance reduced Monte-Carlo simulation of the probability that the portfolio loss exceeds b = 20. The aim is to estimate the probability that the portfolio loss as defined in equation (12) exceeds a given level b for T = 1. First we consider a moderate loss (b = 10) in table 1 and then a high loss (b = 20) in table 2. The Monte-Carlo simulation involves N = 1,000 trajectories and n = 100 discretization steps per trajectory. For the purpose of Gauss-Legendre quadrature, we choose n GL = 64 abscissae and weights 8. For each example (moderate and high loss), we first conduct a benchmark simulation without variance reduction. Then, we conduct a variance reduced simulation employing several constant approximations of the elasticity ɛ. Finally, the approximation of ɛ by Gauss-Legendre quadrature (integration) is considered. As results, we present the estimated probability ˆP (L > b X 0 ), its standard error ˆσ ˆP, the empirical variance ˆσ 2ˆP and the variance ratio ( VarRatio ), i.e. the ratio between the benchmark s empirical variance and the empirical variance of the approach considered. 8 Abscissae and weight values are available on [15]. 10
11 5 Discussion of results The results indicate that the presented approach is mainly suitable for settings in which the aim is to estimate probabilities of high losses. In a standard Monte-Carlo simulation, only few trajectories yield very high losses. Thus, for the simulation of probabilities of high losses, only few trajectories contribute to the Monte-Carlo estimator but increase its empirical variance. The presented importance sampling approach achieves to drag an increased amount of trajectories into the area of interest, i.e. the area of high losses. This result is in accordance with the finding that the same importance sampling approach in the Monte-Carlo simulation of options delivers the best results for deep-out-of-the-money options [11]. The integration method yields better results than the constant approximation approach. Also this finding is intuitive: the integration method accurately resembles the shape of the ɛ-surface. Contrastingly, the constant approach delivers only a very crude approximation of the true elasticity. Also this result is in line with the variance reduced Monte-Carlo simulation of options [11]. Finally, the optimal constant ɛ value depends on the specific parameters of a Monte-Carlo simulation. While for the moderate loss environment (b = 10) a value of ɛ = 7.5 delivers best results, for the high loss environment (b = 20) a value of ɛ = 15 performs best. For too high ɛ values the variance is increased compared to the benchmark. 6 Conclusion The purpose of this article is to introduce an approach that allows to significantly reduce the empirical variance of VaR Monte-Carlo estimators. After laying the theoretical foundation, we derive variance reduced VaR estimators and introduce suitable elasticity approximations as requirement for importance sampling. Numerical results show that this method actually serves to reduce standard errors of estimators, mainly in situations where probabilities of high losses are estimated. The computationally more demanding integration method delivers better results, but also the comparably simple constant approximation method strongly decreases variance. 7 Acknowledgments I would like to thank Prof. Dr. Hermann Singer (FernUniversität in Hagen, Lehrstuhl für angewandte Statistik und Methoden der empirischen Sozialforschung) for lay- 11
12 ing the foundation of my research, his interest in my work, and the supervision of my doctoral project. This work was kindly supported by the Konrad Adenauer Foundation, the donor of my graduate scholarship. References [1] Singer H. Finanzmarktökonometrie. Wirtschaftswissenschaftliche Beiträge Nr Physica-Verl.; [2] Singer H. Importance sampling for Kolmogorov backward equations. AStA Advances in Statistical Analysis. 2014;98(4): [3] Glasserman P, Heidelberger P, Shahabuddin P. Variance reduction techniques for estimating value-at-risk. Management Science. 2000;46(10): [4] Hoogerheide L, van Dijk HK. Bayesian forecasting of value at risk and expected shortfall using adaptive importance sampling. International Journal of Forecasting. 2010;26(2): [5] Hull JC. Options, futures, and other derivatives. Pearson Education; [6] ESMA, EBA, EIOPA, Joint Comittee of the European Supervisory Authorities. Final draft regulatory technical standards with regard to presentation, content, review and provision of the key information document, including the methodologies underpinning the risk, reward and costs information in accordance with Regulation (EU) No 1286/2014 of the European Parliament and of the Council; Last accessed: August 29, 2016, 2:30 p.m. 20Standards/JC%202016%2021%20(Final%20draft%20RTS% 20PRIIPs%20KID%20report).pdf. [7] Baule R, Tallau C. The pricing of path-dependent structured financial retail products: The case of bonus certificates. Journal of Derivatives. 2011;18(4). [8] Baule R. The order flow of discount certificates and issuer pricing behavior. Journal of Banking & Finance. 2011;35(11): [9] Glasserman P. Monte Carlo methods in financial engineering. Applications of mathematics ; 53. New York, NY ; Berlin ; Heidelberg [u.a.]: Springer; [10] Kloeden PE, Platen E. Numerical solution of stochastic differential equations. Berlin ; New York: Springer-Verlag;
13 [11] Müller A. Approximations of option price elasticities for importance sampling. FernUniversität in Hagen Fakultät für Wirtschaftswissenschaft Discussion paper No. 499; [12] Müller A. Varianzreduzierende Verfahren zur Optionspreisbewertung [Master thesis]. FernUniversität in Hagen Fakultät für Wirtschaftswissenschaft; [13] Antia HM. Numerical Methods for Scientists and Engineers. No. Bd. 1 in Numerical Methods for Scientists and Engineers. Springer Verlag NY; [14] Bronstein IN, Semendjajew KA, Musiol G, Mühlig H. Taschenbuch der Mathematik. Verlag Harri Deutsch; [15] Kamermans MP. Gaussian Quadrature Weights and Abscissae; Last accessed: July 27, 2016, 9:21 p.m. bezierinfo/legendre-gauss.html. [16] Flanders H. Differentiation under the integral sign. The American Mathematical Monthly. 1973;80(6):
14 g q! " # # #!! $ " % & $ " $ " ' ( ) * + *, -. / * + *. * ( 5 * + * ( ) * * * ( - 1 ) 6 * / * 2 : ; + < = - ) * ( > 0 ) 1 +? * A B A B B C D E F G H H I I J K L M N L O G F L E P F Q Q K L M R L O I P M N S I E T I F P I E K G N U J I G E U V N L O I Q W X Y L E I P Z P F G H I Q W [ N A \ A B B C ] L ] ^ I P O K U U K J N E K F L F _ ` I L E U K L a F L E I U E U b K E T c L O F M I L F R U d L E P K L U K G e F E K ^ N E K F L D G T Q I J Y E f W A A A B B C e N h K S R S c L E P F J i d L _ I P I L G I _ F P e K h I O a F L E K L R F R U j k K U G P I E I D K L M I P W m I P S N L L N P K N l Q I A n A B B C c K L I m I R P K U E K H _ Y P O N U S I T P O K S I L U K F L N Q I o K L p N G H K L M e N G H W k N L K I Q p P F l Q I S o F P E _ I Q O E W q L O P I N A B B C c h J I G E I O q p F U E I P K F P K c U E K S N E K F L K L V K L N L G K N Q q J J Q K G N E K F L U e N f f F L K W Z T F S N A r A B B C s t u v u w x y s z { } x w ~ } w ~ u ƒ x u v x v z v ˆ y Š } z u x w ~ Œ u y w v { z } Š x u y A A B B C s } u u u } y ~ s z { } x w ~ } z x v { w ~ u v w x v u } x v { Š } z u F Q H I P Ž } w u z w s v } u x v w u } x v z x Ž } w u z w s v } A A B B C k i L N S K G c I G E U F _ ] U T F P K L M D E K I J K G W k I L K U š N M L I P W m I Q S R A C A B B C k I P c K L _ Q R U U ^ F L F U E I L N l b I K G T R L M I L N R _ O N U œ N U T j X Q I K G T M I b K G T E K L I K L I S L K G T E j H F F J I P N E K ^ I L k K U J F L I L E I L j a F L E P F Q Q I P j D J K I Q V N L O I Q W X Y L E I P Z P F G H I Q W [ N A A B B C V N U E q L N Q i E K G ] J E K F L g N Q R N E K F L b K E T X q ` a m e N f f F L K W Z T F S N n B A B B C a F L O K E K F L N Q X N R U U j m I P S K E I V K Q E I P K L M b K E T q J J Q K G N E K F L E F g F Q N E K Q K E i c U E K S N E K F n \ A B B C š I l A ž B N R _ O I S p P Y _ U E N L O Ÿ R P o I b I P E R L M ^ F L d L E I P L I E j L E I P L I T S I n A A B B C I L E P N Q l N L H j F S S R L K H N E K F L R L O V K L N L f U E N l K Q K E E c K L I o I U E N L O U N R _ L N T S n n A B B C X Q F l N Q K f N E K F L N L O q U U I E p P K G I U Ÿ š T K G T Z P N O I j ] U k F a I L E P N Q o N L H U V N G I K L D S N Q Q ] J I L c G F L F S K I A B B C d L E I P L N E K F L N Q p F Q K G i a F F P O K L N E K F L N L O D K S J Q I e F L I E N P i p F Q K G i ` R Q I U D K L M I P W m I P S N L L a T P K U E K N L e N N X F E E T N P O p K I E U G T L Y E E I P W ` F Q _ e F T P W o I L N S K L L Y E E I P W ` F Q _ š N M L I P W m I Q S R E o I P M I P W š F Q _ P N S š N M L I P W m I Q S R n r A B B e N E G T K L M J P F f I U U I N R _ l I P R _ Q K G T I L Z I K Q N P l I K E U S P H E I L D E F J U W e K G T N I Q e N f f F L K W Z T F S N n A B B š N i _ K L O K L M J P F f I U U I K L p N P H U K E R N E K F L I L j I K L I I S J K P K U G T I L N Q i U n A B B c œ Z ` ] p j k ` d g c œ p ] ` Z V ] d ] D c c a Z d ] œ N O F b L U K O I N L O R J U K O I P K U H _ P N S I b F P H V Q K I W D N l K L I Z I E f L I P W D E I _ N L ` O O I P W š K Q T I Q S X N P E L I P W d ^ N L ` K G N P O F ` R O F Q J T W D N L O P n C A B B a F L U R Q E K L M d L G I L E K ^ I U K L a F L E I U E U D G T Q I J Y E f W g F Q H I P
15 @ n A B B q X I L I E K G q Q M F P K E T S _ F P N o K j ] l I G E K ^ I š K L L I P j k I E I P S K L N E K F L p P F l Q I S K L N Z P N L U J F P E N E K F L j p P F G R P I S I L E q R G E K B A B B p N P N Q Q I Q M P I I O i N Q M F P K E T S U _ F P J N G H K L M R L I ª R N Q U J T I P I U K L E F N G R l F K O N Q U E P K J F P N G R l F \ A B B D c e S F O I Q K L M b K E T U K L M R Q N P S F S I L E S N E P K G I U p N P E d Ÿ e j c U E K S N E K F L F _ E K S I U I P K A A B B D c e S F O I Q K L M b K E T U K L M R Q N P S F S I L E S N E P K G I U p N P E d d Ÿ e j c U E K S N E K F L F _ U N S J Q I O U E F G T N U E K G O K I P I L E K N Q I ª R N E K F L U o R I P W Z F l K N U p N L H P N E f W X K U I Q T I P R l N G T W Z K S F o F P E _ I Q O E W q L O P I N U Z K Q Q K W Z T F S N U X I T P K L M W m I P S N L L D K L M I P W m I P S N L L D K L M I P W m I P S N n A B B F L U I L U R N Q I c K f K I L f l I b I P E R L M R L O j ^ I P l I U U I P R L M ` O O I P W š K Q T I Q S L E I P U R G T R L M I L S K E E I Q U O I P k N E N c L ^ I Q F J S I L E q L N Q i U K U ` I R G T I P W c Q S N P «k @ A B B I M N Q L G I P E N K L E i d U m N S F L K f N E K F L F _ N b E T I ` K M T E š N M L I P W m I Q S R E q L U b I P q D T F P E ] ^ I P ^ K r A B B V N U E a F L E K L R F R U j k K U G P I E I k q V j V K Q E I P U e N f f F L K W Z T F S A B \ B R N L E K E N E K ^ I c ^ N Q R K I P R L M ^ F L e R Q E K j I ^ I Q e N P H I E K L M U i U E I S A B \ B R N U K j a F L E K L R F R U e N h K S R S c L E P F J i k K U E P K l R E K F L q J J P F h K S N E K F L b K E T I P L I Q k I L U K C A B \ B D F Q ^ K L M N o K j ] l I G E K ^ I š K L L I P k I E I P S K L N E K F L p P F l Q I S K L N Z P N L U J F P E N E K F L p P F G R P I S I L E q R G E K A B \ B q P I D T F P E Z I P S D E F G H q U U I E ` I E R P L U p P I O K G E N l Q I q L c h E I L O I O c S J K P K G N Q q L N Q i U K r B A B \ B c R P F J K U G T I X I U R L O T I K E U U i U E I S I K S g I P M Q I K G T c K f K I L f S I U U R L M I L ^ F L q H R E H P N L H I L T R U I P L S K E k c q F P I L f W e N P K L N e N f f F L K W Z T F S N U e N f f F L K W Z T F S N U ` I R G T I P W c Q S N P o R I P W Z F l K N U p N L H P N E f W X K U I Q T I P e N f f F L K W Z T F S N U ` I R G T I P W c Q S N P D N P E F P K R U W V P N L r \ A B \ B p N E E I P L U K L ] l I G E j ] P K I L E I O q L N Q i U K U o Q N K S I P W œ K G F Q N U o F P E _ I Q O E W q L O P I N U p N L H P N E f W X K U I Q T I r A A B \ B Z T I R f L I E U j N Q O F P j p R f f Q I N L O œ I R E P N Q a P F U U j a N J K E N Q j d L E I L U K E i D E P R G E R P N Q a T N L M r n A B \ B e F L I E N P i p F Q K G i N L O o F F S j o R U E a i G Q I U Ÿ Z T I ` F Q I F _ a F S S R L K G N E K F A B \ B F L U I L U R N Q I c K f K I L f l I b I P E R L M R L O ^ I P l I U U I P R L M S K E E I Q U k c q ] R E J R E j ^ U ž d L J R E F P K I L E K I P R L r r A B \ B a F L U K U E I L E e F O I Q K L M F _ ` K U H q ^ I P U I o I T N ^ K F P b K E T D J I G E P N Q ` K U H e I N U R P I U D E K I J K G W k I L K U š N M L I P W m I Q S R E L Y E E I P W ` F Q _ š N M L I P W m I Q S R E ` I R G T I P W c Q S N P ` O O I P W š K Q T I Q S š G T E I P W m N L U p I E I P e N f f F L K W Z T F S N U
16 r A B \ B k I P ^ K P E R I Q Q I p I I P c K L I q L b I L O R L M O I P k c q f R P H F L U I L U R N Q I L c K f K I L f j l I b I P E R L M ` I R G T I P W c Q S N r A B \ B s w ƒ w { u ˆ y x v { ˆ } y u } u } Š } w { u u w } x v { y ˆ r C A B \ B s w } u u u } y ~ z { } x w ~ } z x v { w ~ u z w x ƒ x u v x v z w } x ˆ ˆ y x v { ˆ } z u x w ~ { x z z w x v u y w w x v { y v w } x v r A B \ B c ª R K E i N L O c K G K I L G i K L ` I M K F L N Q p R l Q K G X F F O D R J J Q i b K E T d S J I P _ I G E N l F R P e F l K Q K E i m F P K f F L E N Q ^ I P U R U g I P E K G N Q c ª R N Q K f N E K F L } s v Ž } w u z w s v } u Ž } w u z w s v } u v { v v x v u s } v z ± z u B A B \ B q T i l P K O N Q M F P K E T S _ F P E T I G N J N G K E N E I O ^ I T K G Q I P F R E K L M J P F l Q I S b K E T E T P I I j O K S I L U K F L N Q Q F N O K L M G F L U E P N K L E U Ž } w u z w s v } \ A B \ B q E P I I U I N P G T J P F G I O R P I _ F P E T I G F L E N K L I P P I Q F G N E K F L J P F l Q I S } w u } z } x v Ž } w u z w s v } A A B \ \ q O ^ N L G I O ² j c K G K I L G K I U _ F P a a ` j N L O o a a j e F O I Q Q Z F b N P O U p I I P j l N U I O k c q a F L E P F Q Q K L n A B \ \ Z T I c I G E U F _ a I L E P N Q o N L H a F S S R L K G N E K F L F L V K L N L G K N Q D E N l K Q K E i Ÿ q D i U E I S N E K f N E K F L F _ E T I c S J K P K G N Q c ^ K O I L A B \ \ U R L M U H F L f I J E I f R P q Q Q F H N E K F L ^ F L F F J I P N E K F L U ^ F P E I K Q I L K L O I P H F F J I P N E K ^ I L Z P N L U J F P E O K U J F U K E K F r A B \ \ X P I L f I L I K L I P I M K E K S N E K F L U E N N E Q K G T I P e N L N T S I L M I M I L Y l I P P I O K E K L U E K E R E I L f R P g I P T K L O I P R L M ^ F L o N L H I L j R L O š K P E U G T N _ E U H P K U I A B \ \ a F L E P F Q Q K L M K S D E N O E S N P H I E K L M c K L I q L N Q i U I O I U m N M I L I P D G T N R _ I L U E I P b I E E l I b I P l U A B \ A B \ \ q D E P R G E R P N Q q J J P F N G T E F V K L N L G K N Q D E N l K Q K E i Ÿ ] L E T I o I L I _ K G K N Q ` F Q I F _ ` I M R Q N E F P i X F ^ I P L N L G C A B \ \ k N E N c L ^ I Q F J S I L E q L N Q i U K U j D H N Q I L I P E P M I R L O P I R f U H N Q I L I P E P M A B \ \ a F L E P F Q Q K L M F P M N L K U N E F P K U G T I P c L E U G T I K O R L M I L Ÿ F L f I J E K F L I Q Q I l I P Q I M R L M I B A B \ \ } x u v w x u } v { x v x u v w z u x w v { { u v { u v u x v u u ¹ ˆ z } w x u w x u Ž u x ˆ x u z u z { ~ u v } v } w x v Œ ³ u } x z ~ u z Œ u y ~ u } z } L Y E E I P W ` F Q _ e F T P W o I L N S K L š N M L I P W m I Q S R E D E P N L M S I K I P W ` I K L T N P O V K I O Q I P W e N E E T K N U e I P l I G H U W E I V Q K I W D N l K L I o N R I P W N E T N P K L N e F T P W o I L N S K L š N M L I P W m I Q S R E x z ~ u z Œ ³ u } s v } u u z z v x w µ x v v u } z } x v y ~ u } z z x u º x v u z y x s v w» u u ˆ u } u v
17 \ A B \ \ ½ v u ¾ z x w u } x v x v y u u v u v w ~ u w ~ u } w ~ u u z } u w w u u x v } u x y ~ v x u A A B \ \ s w } u u u } y ~ ˆ } y u } u } w ~ u y v w x v u } } u w } x u z ˆ } z u } w u } z } x v Ž } w u z w s v } n A B \ \ s v y w x v z s ˆ ˆ } y ~ w Š } x y x v { ˆ z u ¹ Ž } } x u } ˆ w x v µ µ v A B \ \ Ž z { v ƒ Š } µ u v v u u w u u } v { u z z v } v w w x v } x w v x u } x w } u v ½ u v w x w w u v w x y ~ u } r A B \ \ s } u y w x v ˆ ˆ } y ~ } z x v { w ~ u } u y w v { z u ˆ y x v { } u x v x x µ w x v ˆ } z u Ž } w u z w s v } A B \ \ } u v À v u ˆ z u v w x w ~ Á u w u } { u v u x }  Á t s } u A B \ A w } y w } z ~ v { u Š w w u } v v u u z ˆ u v w  ~ x v x v ˆ } x v s z w u v } { w µ { v u } Á u z C A B \ A u { } x y ~ u Œ x x u v Á u } } u } v { u v à } x v v µ x } w y ~ w z x y ~ u Œ x x v { u u v w x Œ ~ u v u } u w } x u z x y ~ u v s z w u } u } } { v A B \ A Ä Å Æ Ç È É Ê Ê Ë Ì Æ Í Î Ï Ë Ð Ñ Ò ½ v w u } v z x µ x v { ¹ w u } v z x w x u x w ~ Œ Ó x x u v v x } v u v w z x x z x C B A B \ A Ô v x y ~ u ½ v w u } ˆ } u w w x v u v u } z u v } x z u v x v u } C \ A B \ A v w } ˆ x u x u } w u s v z u v ½ v w u } w x v u v x v µ x z u v ¼ u w µ u } u v u } u y À w u v } u s z } u } x u ~ u x Œ v ~ { u v Ž x v y u z z v x w µ s v } u z u x v u s v } u Œ ³ u } x z ~ u z Œ ³ u } x z ~ u z Ž } u v v u } x v x y z v v } x u ~ u z Ž u } { u } z } x º u } } x u } x y C A A B \ n u v w } z Ž v ½ v u ˆ u v u v y u v x v v y x z w x z x w  s z u Š u } u y w Á } v C n A B \ n v u } { { u v u } w x v x w ~ x } u y w u u y ~ v x y z ~ v { u z z u v y ~ t x z u } A B \ n v u w } Š z x y v s u w Š } x y u  ~ u v z u v x v { À ˆ Á x w w ~ u Ö u } u } Ž C r A B \ n ˆ u } v w u v x v µ x z u v ¼ u w µ u } u v u x v u u v w } ˆ x u ˆ w x z u v z C A B \ n w x x { u x w v x w w x v } { v x w x v v Š u } v  t u w z w v { à u } z u { v { u v Ž x u ½ v } w x v u } } u x w v { v w µ C A B ½ v y u v w x u } s v y u s w u u v w u y ~ v z { À v u } ¼ w x v z v ½ v w u } v w x v z Š u } x w } x v { Ž u } { u } z } x º u } } x u } x y ~ Ž } u v v u } x v x y Œ ³ u } x z ~ u z z v v } x u ~ u z z x ~ } x w x v x v v u } z } x v y ~ u } z v } u s z } u Œ v ~ { u v Ž x v y
18 C C A B v x y ~ u x µ x u v µ u u } w v { ³ u v w z x y ~ u } z u x v u s v } u } u x ˆ } w u v w ~ u w u } x w u } w v u z ˆ u v w s v z x Á v v w u u C A B \ r v u v z u Š u u } ƒ ~ z x v u } s ƒ ƒ x µ x u v µ u z z v x w µ s v } u z u v u } w } { Œ u y ~ u } z B A B \ r } ˆ } u v µ x u z z u Œ u { z x u } v { u x v u } µ u x v à ~ } v { v ± u z w ~ ˆ x z z x u x v Ø u } z x y u } x ~ \ A B \ r u z x v { x w ~ x v { z } u v w w } x y u x v { u } Á u } v v Š } w ½ ½ ½  t u w x w x A A B \ r x u w u u } z x y ~ u Ž u } à y x y ~ w x { v { v s u v v { u v à } u x v u u } x v { w u ˆ ~ v Ù Æ Ú Û Ë Ú Ü Ý Þ Ë Ì Î ß É à È Æ Î Ê Ê Ú Ì á Ú Ì Æ Î à â Î È ã Ì Û Î à Î à Š } w ~ u x v u u Ž u } à y x y ~ w x { v { u Á ³ } u } w n A B \ À v { u x ~ u x w u } À v x y ~ u } ~ u x w x v µ x z u v ¼ u w µ u } u v Œ ³ u } x z ~ u z u z z v x w µ s v } u } w v u } A B \ ½ v u w u v w x v ˆ ˆ z x u } ƒ ˆ u y x x y u y v x u y ˆ u x w ~ w v u z t à v w u } x u } u v w u } x y u v x u } u v w ˆ ˆ z x u } y ~ } y w u }  w } y u z v ˆ u y x z x r A B \ s v ˆ ˆ z x y w x v w ~ u ˆ w ƒ y z z ƒ ˆ } x w w } x v y u } u y u à z z u } s } x v v w u ƒ } z ˆ w x v ˆ } x y x v A B \ s» x v w ˆ ˆ z x y w x v w ~ u ˆ w ƒ y z z ƒ ˆ } x w v x ˆ } w v y u à z z u } s } x v ˆ z x v { w } x v y u } u y u ˆ w x v ˆ } x y x v A B \ x z w u ¹ x x u z x ~ } v w x v ƒ x y } u w u x v { u } Á u } v v w w u ˆ y u u z x v { v { u x v ½ ˆ } w v y u ˆ z x v C A B \ s ~ u } s u y w x u v x y w x v z x ~ } x w x A B \ s ˆ ˆ } ¹ x w x v ˆ w x v ˆ } x y u u z w x y x w x u } x ˆ } w v y u à z z u } s } x v ˆ z x v { r B B A B \ ± } x v y u } u y u ± z u w Œ x v w u ƒ } z x z w x v à z z u } s } x v
Framework for functional tree simulation applied to 'golden delicious' apple trees
Purdue University Purdue e-pubs Open Access Theses Theses and Dissertations Spring 2015 Framework for functional tree simulation applied to 'golden delicious' apple trees Marek Fiser Purdue University
More informationIEOR E4703: Monte-Carlo Simulation
IEOR E4703: Monte-Carlo Simulation Output Analysis for Monte-Carlo Martin Haugh Department of Industrial Engineering and Operations Research Columbia University Email: martin.b.haugh@gmail.com Output Analysis
More informationVariance Reduction Techniques for Monte Carlo Simulations with Stochastic Volatility Models
Variance Reduction Techniques for Monte Carlo Simulations with Stochastic Volatility Models Jean-Pierre Fouque North Carolina State University SAMSI November 3, 5 1 References: Variance Reduction for Monte
More informationOptimal Control of PDEs
Optimal Control of PDEs Suzanne Lenhart University of Tennessee, Knoville Department of Mathematics Lecture1 p.1/36 Outline 1. Idea of diffusion PDE 2. Motivating Eample 3. Big picture of optimal control
More information3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 REDCLIFF MUNICIPAL PLANNING COMMISSION FOR COMMENT/DISCUSSION DATE: TOPIC: April 27 th, 2018 Bylaw 1860/2018, proposed amendments to the Land Use Bylaw regarding cannabis
More informationLA PRISE DE CALAIS. çoys, çoys, har - dis. çoys, dis. tons, mantz, tons, Gas. c est. à ce. C est à ce. coup, c est à ce
> ƒ? @ Z [ \ _ ' µ `. l 1 2 3 z Æ Ñ 6 = Ð l sl (~131 1606) rn % & +, l r s s, r 7 nr ss r r s s s, r s, r! " # $ s s ( ) r * s, / 0 s, r 4 r r 9;: < 10 r mnz, rz, r ns, 1 s ; j;k ns, q r s { } ~ l r mnz,
More informationAn Introduction to Optimal Control Applied to Disease Models
An Introduction to Optimal Control Applied to Disease Models Suzanne Lenhart University of Tennessee, Knoxville Departments of Mathematics Lecture1 p.1/37 Example Number of cancer cells at time (exponential
More informationIntroduction to the Numerical Solution of SDEs Selected topics
0 / 23 Introduction to the Numerical Solution of SDEs Selected topics Andreas Rößler Summer School on Numerical Methods for Stochastic Differential Equations at Vienna University of Technology, Austria
More informationKolmogorov Equations and Markov Processes
Kolmogorov Equations and Markov Processes May 3, 013 1 Transition measures and functions Consider a stochastic process {X(t)} t 0 whose state space is a product of intervals contained in R n. We define
More informationDiscretization of SDEs: Euler Methods and Beyond
Discretization of SDEs: Euler Methods and Beyond 09-26-2006 / PRisMa 2006 Workshop Outline Introduction 1 Introduction Motivation Stochastic Differential Equations 2 The Time Discretization of SDEs Monte-Carlo
More informationPrincipal Secretary to Government Haryana, Town & Country Planning Department, Haryana, Chandigarh.
1 From To Principal Secretary to Government Haryana, Town & Country Planning Department, Haryana, Chandigarh. The Director General, Town & Country Planning Department, Haryana, Chandigarh. Memo No. Misc-2339
More informationChapter 4: Monte-Carlo Methods
Chapter 4: Monte-Carlo Methods A Monte-Carlo method is a technique for the numerical realization of a stochastic process by means of normally distributed random variables. In financial mathematics, it
More informationIntroduction to Algorithmic Trading Strategies Lecture 10
Introduction to Algorithmic Trading Strategies Lecture 10 Risk Management Haksun Li haksun.li@numericalmethod.com www.numericalmethod.com Outline Value at Risk (VaR) Extreme Value Theory (EVT) References
More information! " # $! % & '! , ) ( + - (. ) ( ) * + / 0 1 2 3 0 / 4 5 / 6 0 ; 8 7 < = 7 > 8 7 8 9 : Œ Š ž P P h ˆ Š ˆ Œ ˆ Š ˆ Ž Ž Ý Ü Ý Ü Ý Ž Ý ê ç è ± ¹ ¼ ¹ ä ± ¹ w ç ¹ è ¼ è Œ ¹ ± ¹ è ¹ è ä ç w ¹ ã ¼ ¹ ä ¹ ¼ ¹ ±
More information1. Allstate Group Critical Illness Claim Form: When filing Critical Illness claim, please be sure to include the following:
Dear Policyholder, Per your recent request, enclosed you will find the following forms: 1. Allstate Group Critical Illness Claim Form: When filing Critical Illness claim, please be sure to include the
More informationprobability of k samples out of J fall in R.
Nonparametric Techniques for Density Estimation (DHS Ch. 4) n Introduction n Estimation Procedure n Parzen Window Estimation n Parzen Window Example n K n -Nearest Neighbor Estimation Introduction Suppose
More informationEstimation of Operational Risk Capital Charge under Parameter Uncertainty
Estimation of Operational Risk Capital Charge under Parameter Uncertainty Pavel V. Shevchenko Principal Research Scientist, CSIRO Mathematical and Information Sciences, Sydney, Locked Bag 17, North Ryde,
More informationAppendix of the paper: Are interest rate options important for the assessment of interest rate risk?
Appendix of the paper: Are interest rate options important for the assessment of interest rate risk? Caio Almeida,a, José Vicente b a Graduate School of Economics, Getulio Vargas Foundation b Research
More information5 December 2016 MAA136 Researcher presentation. Anatoliy Malyarenko. Topics for Bachelor and Master Theses. Anatoliy Malyarenko
5 December 216 MAA136 Researcher presentation 1 schemes The main problem of financial engineering: calculate E [f(x t (x))], where {X t (x): t T } is the solution of the system of X t (x) = x + Ṽ (X s
More informationLecture 4: Numerical Solution of SDEs, Itô Taylor Series, Gaussian Process Approximations
Lecture 4: Numerical Solution of SDEs, Itô Taylor Series, Gaussian Process Approximations Simo Särkkä Aalto University Tampere University of Technology Lappeenranta University of Technology Finland November
More informationModel Specification Testing in Nonparametric and Semiparametric Time Series Econometrics. Jiti Gao
Model Specification Testing in Nonparametric and Semiparametric Time Series Econometrics Jiti Gao Department of Statistics School of Mathematics and Statistics The University of Western Australia Crawley
More informationSFB 823. A simple and focused backtest of value at risk. Discussion Paper. Walter Krämer, Dominik Wied
SFB 823 A simple and focused backtest of value at risk Discussion Paper Walter Krämer, Dominik Wied Nr. 17/2015 A simple and focused backtest of value at risk 1 by Walter Krämer and Dominik Wied Fakultät
More information1 Introduction. 2 Diffusion equation and central limit theorem. The content of these notes is also covered by chapter 3 section B of [1].
1 Introduction The content of these notes is also covered by chapter 3 section B of [1]. Diffusion equation and central limit theorem Consider a sequence {ξ i } i=1 i.i.d. ξ i = d ξ with ξ : Ω { Dx, 0,
More informationLecture 4: Numerical Solution of SDEs, Itô Taylor Series, Gaussian Approximations
Lecture 4: Numerical Solution of SDEs, Itô Taylor Series, Gaussian Approximations Simo Särkkä Aalto University, Finland November 18, 2014 Simo Särkkä (Aalto) Lecture 4: Numerical Solution of SDEs November
More informationExperience Rating in General Insurance by Credibility Estimation
Experience Rating in General Insurance by Credibility Estimation Xian Zhou Department of Applied Finance and Actuarial Studies Macquarie University, Sydney, Australia Abstract This work presents a new
More informationA simple nonparametric test for structural change in joint tail probabilities SFB 823. Discussion Paper. Walter Krämer, Maarten van Kampen
SFB 823 A simple nonparametric test for structural change in joint tail probabilities Discussion Paper Walter Krämer, Maarten van Kampen Nr. 4/2009 A simple nonparametric test for structural change in
More informationLecture 4: Introduction to stochastic processes and stochastic calculus
Lecture 4: Introduction to stochastic processes and stochastic calculus Cédric Archambeau Centre for Computational Statistics and Machine Learning Department of Computer Science University College London
More informationQUESTIONS ON QUARKONIUM PRODUCTION IN NUCLEAR COLLISIONS
International Workshop Quarkonium Working Group QUESTIONS ON QUARKONIUM PRODUCTION IN NUCLEAR COLLISIONS ALBERTO POLLERI TU München and ECT* Trento CERN - November 2002 Outline What do we know for sure?
More informationThe University of Bath School of Management is one of the oldest established management schools in Britain. It enjoys an international reputation for
The University of Bath School of Management is one of the oldest established management schools in Britain. It enjoys an international reputation for the quality of its teaching and research. Its mission
More informationVector analysis. 1 Scalars and vectors. Fields. Coordinate systems 1. 2 The operator The gradient, divergence, curl, and Laplacian...
Vector analysis Abstract These notes present some background material on vector analysis. Except for the material related to proving vector identities (including Einstein s summation convention and the
More informationGaussian, Markov and stationary processes
Gaussian, Markov and stationary processes Gonzalo Mateos Dept. of ECE and Goergen Institute for Data Science University of Rochester gmateosb@ece.rochester.edu http://www.ece.rochester.edu/~gmateosb/ November
More informationNumerical Integration of SDEs: A Short Tutorial
Numerical Integration of SDEs: A Short Tutorial Thomas Schaffter January 19, 010 1 Introduction 1.1 Itô and Stratonovich SDEs 1-dimensional stochastic differentiable equation (SDE) is given by [6, 7] dx
More informationMonte-Carlo MMD-MA, Université Paris-Dauphine. Xiaolu Tan
Monte-Carlo MMD-MA, Université Paris-Dauphine Xiaolu Tan tan@ceremade.dauphine.fr Septembre 2015 Contents 1 Introduction 1 1.1 The principle.................................. 1 1.2 The error analysis
More informationStochastic Volatility and Correction to the Heat Equation
Stochastic Volatility and Correction to the Heat Equation Jean-Pierre Fouque, George Papanicolaou and Ronnie Sircar Abstract. From a probabilist s point of view the Twentieth Century has been a century
More informationParametric Techniques Lecture 3
Parametric Techniques Lecture 3 Jason Corso SUNY at Buffalo 22 January 2009 J. Corso (SUNY at Buffalo) Parametric Techniques Lecture 3 22 January 2009 1 / 39 Introduction In Lecture 2, we learned how to
More informationIntroduction to numerical simulations for Stochastic ODEs
Introduction to numerical simulations for Stochastic ODEs Xingye Kan Illinois Institute of Technology Department of Applied Mathematics Chicago, IL 60616 August 9, 2010 Outline 1 Preliminaries 2 Numerical
More informationSymbols and dingbats. A 41 Α a 61 α À K cb ➋ à esc. Á g e7 á esc. Â e e5 â. Ã L cc ➌ ã esc ~ Ä esc : ä esc : Å esc * å esc *
Note: Although every effort ws tken to get complete nd ccurte tble, the uhtor cn not be held responsible for ny errors. Vrious sources hd to be consulted nd MIF hd to be exmined to get s much informtion
More informationParametric Techniques
Parametric Techniques Jason J. Corso SUNY at Buffalo J. Corso (SUNY at Buffalo) Parametric Techniques 1 / 39 Introduction When covering Bayesian Decision Theory, we assumed the full probabilistic structure
More informationQuantization of stochastic processes with applications on Euler-Maruyama schemes
UPTEC F15 063 Examensarbete 30 hp Oktober 2015 Quantization of stochastic processes with applications on Euler-Maruyama schemes Viktor Edward Abstract Quantization of stochastic processes with applications
More information1. Stochastic Processes and filtrations
1. Stochastic Processes and 1. Stoch. pr., A stochastic process (X t ) t T is a collection of random variables on (Ω, F) with values in a measurable space (S, S), i.e., for all t, In our case X t : Ω S
More informationSome Terminology and Concepts that We will Use, But Not Emphasize (Section 6.2)
Some Terminology and Concepts that We will Use, But Not Emphasize (Section 6.2) Statistical analysis is based on probability theory. The fundamental object in probability theory is a probability space,
More informationUNIVERSITÄT POTSDAM Institut für Mathematik
UNIVERSITÄT POTSDAM Institut für Mathematik Testing the Acceleration Function in Life Time Models Hannelore Liero Matthias Liero Mathematische Statistik und Wahrscheinlichkeitstheorie Universität Potsdam
More informationLecture on Parameter Estimation for Stochastic Differential Equations. Erik Lindström
Lecture on Parameter Estimation for Stochastic Differential Equations Erik Lindström Recap We are interested in the parameters θ in the Stochastic Integral Equations X(t) = X(0) + t 0 µ θ (s, X(s))ds +
More informationMarket Risk. MFM Practitioner Module: Quantitiative Risk Management. John Dodson. February 8, Market Risk. John Dodson.
MFM Practitioner Module: Quantitiative Risk Management February 8, 2017 This week s material ties together our discussion going back to the beginning of the fall term about risk measures based on the (single-period)
More informationMalliavin Calculus in Finance
Malliavin Calculus in Finance Peter K. Friz 1 Greeks and the logarithmic derivative trick Model an underlying assent by a Markov process with values in R m with dynamics described by the SDE dx t = b(x
More informationDifferentiation and Integration
Differentiation and Integration (Lectures on Numerical Analysis for Economists II) Jesús Fernández-Villaverde 1 and Pablo Guerrón 2 February 12, 2018 1 University of Pennsylvania 2 Boston College Motivation
More informationDerivation of Itô SDE and Relationship to ODE and CTMC Models
Derivation of Itô SDE and Relationship to ODE and CTMC Models Biomathematics II April 23, 2015 Linda J. S. Allen Texas Tech University TTU 1 Euler-Maruyama Method for Numerical Solution of an Itô SDE dx(t)
More informationVectors. Teaching Learning Point. Ç, where OP. l m n
Vectors 9 Teaching Learning Point l A quantity that has magnitude as well as direction is called is called a vector. l A directed line segment represents a vector and is denoted y AB Å or a Æ. l Position
More informationDuration-Based Volatility Estimation
A Dual Approach to RV Torben G. Andersen, Northwestern University Dobrislav Dobrev, Federal Reserve Board of Governors Ernst Schaumburg, Northwestern Univeristy CHICAGO-ARGONNE INSTITUTE ON COMPUTATIONAL
More informationHochdimensionale Integration
Oliver Ernst Institut für Numerische Mathematik und Optimierung Hochdimensionale Integration 14-tägige Vorlesung im Wintersemester 2010/11 im Rahmen des Moduls Ausgewählte Kapitel der Numerik Contents
More informationA variational radial basis function approximation for diffusion processes
A variational radial basis function approximation for diffusion processes Michail D. Vrettas, Dan Cornford and Yuan Shen Aston University - Neural Computing Research Group Aston Triangle, Birmingham B4
More informationSolution of Stochastic Optimal Control Problems and Financial Applications
Journal of Mathematical Extension Vol. 11, No. 4, (2017), 27-44 ISSN: 1735-8299 URL: http://www.ijmex.com Solution of Stochastic Optimal Control Problems and Financial Applications 2 Mat B. Kafash 1 Faculty
More information1. CALL TO ORDER 2. ROLL CALL 3. PLEDGE OF ALLEGIANCE 4. ORAL COMMUNICATIONS (NON-ACTION ITEM) 5. APPROVAL OF AGENDA 6. EWA SEJPA INTEGRATION PROPOSAL
SPECIAL MEETING AGENDA SAN ELIJO JOINT POWERS AUTHORITY (SEJPA) TUESDAY JULY 12, 2016 AT 8:30 AM ENCINA WATER POLLUTION CONTROL FACILITY (EWA) BOARD MEETING ROOM 6200 AVENIDA ENCINAS CARLSBAD, CALIFORNIA
More informationOptimal portfolio strategies under partial information with expert opinions
1 / 35 Optimal portfolio strategies under partial information with expert opinions Ralf Wunderlich Brandenburg University of Technology Cottbus, Germany Joint work with Rüdiger Frey Research Seminar WU
More informationAsymptotic distribution of the sample average value-at-risk
Asymptotic distribution of the sample average value-at-risk Stoyan V. Stoyanov Svetlozar T. Rachev September 3, 7 Abstract In this paper, we prove a result for the asymptotic distribution of the sample
More informationExamination paper for TFY4240 Electromagnetic theory
Department of Physics Examination paper for TFY4240 Electromagnetic theory Academic contact during examination: Associate Professor John Ove Fjærestad Phone: 97 94 00 36 Examination date: 16 December 2015
More informationConditional Gauss Hermite Filtering with Application to Volatility Estimation
Conditional Gauss Hermite Filtering with Application to Volatility Estimation Hermann Singer FernUniversität in Hagen October 1, 8 Abstract The conditional Gauss Hermite filter (CGHF) utilizes a decomposition
More informationApproximation of Top Lyapunov Exponent of Stochastic Delayed Turning Model Using Fokker-Planck Approach
Approximation of Top Lyapunov Exponent of Stochastic Delayed Turning Model Using Fokker-Planck Approach Henrik T. Sykora, Walter V. Wedig, Daniel Bachrathy and Gabor Stepan Department of Applied Mechanics,
More informationPerformance Evaluation of Generalized Polynomial Chaos
Performance Evaluation of Generalized Polynomial Chaos Dongbin Xiu, Didier Lucor, C.-H. Su, and George Em Karniadakis 1 Division of Applied Mathematics, Brown University, Providence, RI 02912, USA, gk@dam.brown.edu
More informationSECTION A - EMPLOYEE DETAILS Termination date Employee number Contract/ Department SECTION C - STANDARD DOCUMENTS REQUIRED FOR ALL TERMINATIONS
UNITRANS SA TERMINATION CHECKLIST NOTES: 1. This form applies to all terminations and needs to be submitted along with the required supplementary documentation to the HR Department within the following
More informationDUBLIN CITY UNIVERSITY
DUBLIN CITY UNIVERSITY SAMPLE EXAMINATIONS 2017/2018 MODULE: QUALIFICATIONS: Simulation for Finance MS455 B.Sc. Actuarial Mathematics ACM B.Sc. Financial Mathematics FIM YEAR OF STUDY: 4 EXAMINERS: Mr
More informationControlled Diffusions and Hamilton-Jacobi Bellman Equations
Controlled Diffusions and Hamilton-Jacobi Bellman Equations Emo Todorov Applied Mathematics and Computer Science & Engineering University of Washington Winter 2014 Emo Todorov (UW) AMATH/CSE 579, Winter
More informationA new approach for investment performance measurement. 3rd WCMF, Santa Barbara November 2009
A new approach for investment performance measurement 3rd WCMF, Santa Barbara November 2009 Thaleia Zariphopoulou University of Oxford, Oxford-Man Institute and The University of Texas at Austin 1 Performance
More informationCoherent Systems of Components with Multivariate Phase Type Life Distributions
!#"%$ & ' ")( * +!-,#. /10 24353768:9 ;=A@CBD@CEGF4HJI?HKFL@CM H < N OPc_dHe@ F]IfR@ ZgWhNe@ iqwjhkf]bjwlkyaf]w
More informationarxiv:cond-mat/ v2 [cond-mat.stat-mech] 23 Feb 1998
Multiplicative processes and power laws arxiv:cond-mat/978231v2 [cond-mat.stat-mech] 23 Feb 1998 Didier Sornette 1,2 1 Laboratoire de Physique de la Matière Condensée, CNRS UMR6632 Université des Sciences,
More information3. Linear Regression With a Single Regressor
3. Linear Regression With a Single Regressor Econometrics: (I) Application of statistical methods in empirical research Testing economic theory with real-world data (data analysis) 56 Econometrics: (II)
More informationDoes k-th Moment Exist?
Does k-th Moment Exist? Hitomi, K. 1 and Y. Nishiyama 2 1 Kyoto Institute of Technology, Japan 2 Institute of Economic Research, Kyoto University, Japan Email: hitomi@kit.ac.jp Keywords: Existence of moments,
More informationOC330C. Wiring Diagram. Recommended PKH- P35 / P50 GALH PKA- RP35 / RP50. Remarks (Drawing No.) No. Parts No. Parts Name Specifications
G G " # $ % & " ' ( ) $ * " # $ % & " ( + ) $ * " # C % " ' ( ) $ * C " # C % " ( + ) $ * C D ; E @ F @ 9 = H I J ; @ = : @ A > B ; : K 9 L 9 M N O D K P D N O Q P D R S > T ; U V > = : W X Y J > E ; Z
More informationMean-square Stability Analysis of an Extended Euler-Maruyama Method for a System of Stochastic Differential Equations
Mean-square Stability Analysis of an Extended Euler-Maruyama Method for a System of Stochastic Differential Equations Ram Sharan Adhikari Assistant Professor Of Mathematics Rogers State University Mathematical
More informationProbabilities & Statistics Revision
Probabilities & Statistics Revision Christopher Ting Christopher Ting http://www.mysmu.edu/faculty/christophert/ : christopherting@smu.edu.sg : 6828 0364 : LKCSB 5036 January 6, 2017 Christopher Ting QF
More informationShort-time expansions for close-to-the-money options under a Lévy jump model with stochastic volatility
Short-time expansions for close-to-the-money options under a Lévy jump model with stochastic volatility José Enrique Figueroa-López 1 1 Department of Statistics Purdue University Statistics, Jump Processes,
More informationThe choice of weights in kernel regression estimation
Biometrika (1990), 77, 2, pp. 377-81 Printed in Great Britain The choice of weights in kernel regression estimation BY THEO GASSER Zentralinstitut fur Seelische Gesundheit, J5, P.O.B. 5970, 6800 Mannheim
More informationNew method for solving nonlinear sum of ratios problem based on simplicial bisection
V Ù â ð f 33 3 Vol33, No3 2013 3 Systems Engineering Theory & Practice Mar, 2013 : 1000-6788(2013)03-0742-06 : O2112!"#$%&')(*)+),-))/0)1)23)45 : A 687:9 1, ;:= 2 (1?@ACBEDCFHCFEIJKLCFFM, NCO 453007;
More informationFinancial Econometrics and Volatility Models Extreme Value Theory
Financial Econometrics and Volatility Models Extreme Value Theory Eric Zivot May 3, 2010 1 Lecture Outline Modeling Maxima and Worst Cases The Generalized Extreme Value Distribution Modeling Extremes Over
More informationLecture 11: Regression Methods I (Linear Regression)
Lecture 11: Regression Methods I (Linear Regression) Fall, 2017 1 / 40 Outline Linear Model Introduction 1 Regression: Supervised Learning with Continuous Responses 2 Linear Models and Multiple Linear
More informationELEMENTS OF PROBABILITY THEORY
ELEMENTS OF PROBABILITY THEORY Elements of Probability Theory A collection of subsets of a set Ω is called a σ algebra if it contains Ω and is closed under the operations of taking complements and countable
More informationIntroduction. Stochastic Processes. Will Penny. Stochastic Differential Equations. Stochastic Chain Rule. Expectations.
19th May 2011 Chain Introduction We will Show the relation between stochastic differential equations, Gaussian processes and methods This gives us a formal way of deriving equations for the activity of
More informationMATH4210 Financial Mathematics ( ) Tutorial 7
MATH40 Financial Mathematics (05-06) Tutorial 7 Review of some basic Probability: The triple (Ω, F, P) is called a probability space, where Ω denotes the sample space and F is the set of event (σ algebra
More informationPoisson Jumps in Credit Risk Modeling: a Partial Integro-differential Equation Formulation
Poisson Jumps in Credit Risk Modeling: a Partial Integro-differential Equation Formulation Jingyi Zhu Department of Mathematics University of Utah zhu@math.utah.edu Collaborator: Marco Avellaneda (Courant
More information6. Brownian Motion. Q(A) = P [ ω : x(, ω) A )
6. Brownian Motion. stochastic process can be thought of in one of many equivalent ways. We can begin with an underlying probability space (Ω, Σ, P) and a real valued stochastic process can be defined
More informationLECTURE NOTE #3 PROF. ALAN YUILLE
LECTURE NOTE #3 PROF. ALAN YUILLE 1. Three Topics (1) Precision and Recall Curves. Receiver Operating Characteristic Curves (ROC). What to do if we do not fix the loss function? (2) The Curse of Dimensionality.
More informationIntroduction to Computational Finance and Financial Econometrics Probability Theory Review: Part 2
Introduction to Computational Finance and Financial Econometrics Probability Theory Review: Part 2 Eric Zivot July 7, 2014 Bivariate Probability Distribution Example - Two discrete rv s and Bivariate pdf
More informationBrownian Motion. An Undergraduate Introduction to Financial Mathematics. J. Robert Buchanan. J. Robert Buchanan Brownian Motion
Brownian Motion An Undergraduate Introduction to Financial Mathematics J. Robert Buchanan 2010 Background We have already seen that the limiting behavior of a discrete random walk yields a derivation of
More informationExtreme Value Analysis and Spatial Extremes
Extreme Value Analysis and Department of Statistics Purdue University 11/07/2013 Outline Motivation 1 Motivation 2 Extreme Value Theorem and 3 Bayesian Hierarchical Models Copula Models Max-stable Models
More informationBayesian Inference for DSGE Models. Lawrence J. Christiano
Bayesian Inference for DSGE Models Lawrence J. Christiano Outline State space-observer form. convenient for model estimation and many other things. Bayesian inference Bayes rule. Monte Carlo integation.
More informationFinancial Econometrics and Quantitative Risk Managenent Return Properties
Financial Econometrics and Quantitative Risk Managenent Return Properties Eric Zivot Updated: April 1, 2013 Lecture Outline Course introduction Return definitions Empirical properties of returns Reading
More informationProceedings of the 2014 Winter Simulation Conference A. Tolk, S. Y. Diallo, I. O. Ryzhov, L. Yilmaz, S. Buckley, and J. A. Miller, eds.
Proceedings of the 4 Winter Simulation Conference A. Tolk, S. Y. Diallo,. O. Ryzhov, L. Yilmaz, S. Buckley, and J. A. Miller, eds. EFFCENT STRATFED SAMPLNG MPLEMENTATONS N MULTRESPONSE SMULATON İsmail
More informationAN IDENTIFICATION ALGORITHM FOR ARMAX SYSTEMS
AN IDENTIFICATION ALGORITHM FOR ARMAX SYSTEMS First the X, then the AR, finally the MA Jan C. Willems, K.U. Leuven Workshop on Observation and Estimation Ben Gurion University, July 3, 2004 p./2 Joint
More informationLecture 2: From Linear Regression to Kalman Filter and Beyond
Lecture 2: From Linear Regression to Kalman Filter and Beyond January 18, 2017 Contents 1 Batch and Recursive Estimation 2 Towards Bayesian Filtering 3 Kalman Filter and Bayesian Filtering and Smoothing
More informationLEAD IN DRINKING WATER REPORT
181 US Hwy 46 Mine Hill, NJ 07803 (908) 654-8068 (800) 783-0567 Fax 908-654-8069 LEAD IN DRINKING WATER REPORT Performed At: Montrose School 356 Clark St., South Orange, NJ 07079 Performed For: South Orange
More informationETIKA V PROFESII PSYCHOLÓGA
P r a ž s k á v y s o k á š k o l a p s y c h o s o c i á l n í c h s t u d i í ETIKA V PROFESII PSYCHOLÓGA N a t á l i a S l o b o d n í k o v á v e d ú c i p r á c e : P h D r. M a r t i n S t r o u
More informationOptimal data-independent point locations for RBF interpolation
Optimal dataindependent point locations for RF interpolation S De Marchi, R Schaback and H Wendland Università di Verona (Italy), Universität Göttingen (Germany) Metodi di Approssimazione: lezione dell
More informationEstimation of Quantiles
9 Estimation of Quantiles The notion of quantiles was introduced in Section 3.2: recall that a quantile x α for an r.v. X is a constant such that P(X x α )=1 α. (9.1) In this chapter we examine quantiles
More informationFall 2017 STAT 532 Homework Peter Hoff. 1. Let P be a probability measure on a collection of sets A.
1. Let P be a probability measure on a collection of sets A. (a) For each n N, let H n be a set in A such that H n H n+1. Show that P (H n ) monotonically converges to P ( k=1 H k) as n. (b) For each n
More informationContinuing Education and Workforce Training Course Schedule Winter / Spring 2010
Continuing Education and Workforce Training Course Schedule Winter / Spring 2010 www.southeastmn.edu/training Creating the Solution that Fits! Information Information Registration Registration via our
More informationSequential Monte Carlo Methods for Bayesian Computation
Sequential Monte Carlo Methods for Bayesian Computation A. Doucet Kyoto Sept. 2012 A. Doucet (MLSS Sept. 2012) Sept. 2012 1 / 136 Motivating Example 1: Generic Bayesian Model Let X be a vector parameter
More informationA MODEL FOR THE LONG-TERM OPTIMAL CAPACITY LEVEL OF AN INVESTMENT PROJECT
A MODEL FOR HE LONG-ERM OPIMAL CAPACIY LEVEL OF AN INVESMEN PROJEC ARNE LØKKA AND MIHAIL ZERVOS Abstract. We consider an investment project that produces a single commodity. he project s operation yields
More informationOptimization of a parallel 3d-FFT with non-blocking collective operations
Optimization of a parallel 3d-FFT with non-blocking collective operations Chair of Computer Architecture Technical University of Chemnitz Département de Physique Théorique et Appliquée Commissariat à l
More informationErnesto Mordecki 1. Lecture III. PASI - Guanajuato - June 2010
Optimal stopping for Hunt and Lévy processes Ernesto Mordecki 1 Lecture III. PASI - Guanajuato - June 2010 1Joint work with Paavo Salminen (Åbo, Finland) 1 Plan of the talk 1. Motivation: from Finance
More informationAccurate approximation of stochastic differential equations
Accurate approximation of stochastic differential equations Simon J.A. Malham and Anke Wiese (Heriot Watt University, Edinburgh) Birmingham: 6th February 29 Stochastic differential equations dy t = V (y
More information