Fin285a:Computer Simulations and Risk Assessment Section 6.2 Extreme Value Theory Daníelson, 9 (skim), skip 9.5

Size: px
Start display at page:

Download "Fin285a:Computer Simulations and Risk Assessment Section 6.2 Extreme Value Theory Daníelson, 9 (skim), skip 9.5"

Transcription

1 Fin285a:Computer Simulations and Risk Assessment Section 6.2 Extreme Value Theory Daníelson, 9 (skim), skip 9.5

2 Overview Extreme value distributions Generalized Pareto distributions Tail shapes Using power laws with returns Time aggregation Chebyshev s inequality Summary Fall 2016: LeBaron Fin285a: / 42

3 Extreme value distributions Generalized Pareto distributions Tail shapes Using power laws with returns Time aggregation Chebyshev s inequality Extreme value distributions Summary Fall 2016: LeBaron Fin285a: / 42

4 Extreme value distributions R t daily returns x m,t = max(r t ) over month x m,t follows one of the following Gumbel (exponential decay - like Normal) Frechet (power law decay - fat tail) Wiebull (finite endpoint) Sort of like a central limit theorem Block maxima Key application: floods Fall 2016: LeBaron Fin285a: / 42

5 Extreme value distributions Generalized Pareto distributions Example: fitgpareto.m Definition: Empirical CDF Example: fitgpareto.m Example: fitgpareto.m Tail shapes Generalized Pareto distributions Using power laws with returns Time aggregation Chebyshev s inequality Summary Fall 2016: LeBaron Fin285a: / 42

6 Peaks over thresholds (POT) Distributions for all values exceeding some threshold Converge to the Generalized Pareto Distribution (GPD) Often more useful than distributions for maximums Fall 2016: LeBaron Fin285a: / 42

7 Conditional tail distributions Note: This deals with right tails. F u (x) = Pr(X u x X > u) (6.2.1) As u the distribution of tail values approaches a Generalized Pareto Distribution (GPD). F u (x) = G ξ,β (x) = { 1 (1+ξ x β ) 1 ξ ξ 0 1 e (x β ) ξ = 0 (6.2.2) Fall 2016: LeBaron Fin285a: / 42

8 Fitting a tail for stock returns Matlab has several functions to deal with Generalized Pareto Distributions. What should we do? 1.Adjust data Flip sign (around mean) Find threshold level t = 0.05 or other small level (tailprob in code) Threshold: u = quantile(ret, 1-t) Get right tail ret(ret>u)-u 2.Fit pareto parameters gpfit() Fall 2016: LeBaron Fin285a: / 42

9 Finding extreme quantiles in tails 1.Choose small p,pr(x > x) = p 2.Convert this to probability in tail Pr(X > x X > u) = Pr((X > x)&(x > u)) Pr(X > u) (6.2.3) Pr((X > x)&(x > u)) = Pr(X > x X > u)pr(x > u) (6.2.4) 3.Since we are only interested in x > u, Pr(X > x) = Pr(X > x X > u)pr(x > u) (6.2.5) and Pr(X > x X > u) = Pr(X > x) Pr(X > u) (6.2.6) Fall 2016: LeBaron Fin285a: / 42

10 Conditional CDF Pr(X > x X > u) = Pr(X x X > u) = 1 Pr(X > x) Pr(X > u) Pr(X > x) Pr(X > u) (6.2.7) Fall 2016: LeBaron Fin285a: / 42

11 Picture for a basic tail region t = 0.05,u = Define tail region as starting in the upper 0.05 of the distribution Assume normal distribution (just for example) For a normal, this would start at Pr(X > 1.64) = 0.05 Fall 2016: LeBaron Fin285a: / 42

12 Picture for a basic tail region t = 0.05,p = Now we look at the point where Pr(X > x) = 0.02 For a normal this is x = 2.05 Would like to get this probability conditional on being > 1.64 = u Fall 2016: LeBaron Fin285a: / 42

13 Picture for a basic tail region t = 0.05,p = Pr(X > 2.05 X > 1.64) = Pr(X > 2.05) Pr(X > 1.64) = = 0.4 (6.2.8) The upper 0.02 tail of the entire distribution is the upper 0.4 tail of the conditional distribution beyond the u threshold. Fall 2016: LeBaron Fin285a: / 42

14 Now convert to quantiles in this example Pr(X > 2.05 X > 1.64) = = 0.4 Pr(X 2.05 X > 1.64) = 1 Pr(X > 2.05 X > 1.64) = = 0.6 Now here s what we can do: 1.Adjust the desired tail probability and subtract it from 1, α = 1 p t 2.Use this to find the q α quantile on the GPD distribution using the inverse CDF (matlab knows this function) 3.Add this to the threshold, u. 4.This will be our extreme return at probability p. 5.Finally, flip the sign around the mean. On to matlab... Fall 2016: LeBaron Fin285a: / 42

15 Data transformations The Generalized Pareto distribution starts at zero, and only covers positive numbers It can only give you a CDF of Pr(X x) for x 0 Since we are interested in negative returns we need to transform them twice 1.Flip them around the mean, R t = (R t R t ), Rt = mean(r t ) (6.2.9) 2.Then subtract u, ˆR t = R t u Fall 2016: LeBaron Fin285a: / 42

16 Example: fitgpareto.m First, manipulate returns, and estimate Pareto params, % threshold probability level tailprob = 0.05; % this is t in the notes % Reverse sign on returns and remove the mean % This lets us look at the right tail mret = mean(ret1); adjustret = -1*(ret1-mret); % u = quantile(ret,1-t) threshold return level uthresh = quantile(adjustret,1-tailprob); % u in notes % limit to only returns beyond threshold tailret = adjustret(adjustret>uthresh); % take off threshold so low value is zero tailret = tailret-uthresh; % estimate pareto params pparams = gpfit(tailret); Fall 2016: LeBaron Fin285a: / 42

17 Definition: Empirical CDF CDF for actual data. F e (x) = Pr(x i x) = #x i x N Matlab example: % find the fraction of xref<=y datacdf = empcdf(y,xref); % y can be a vector % often use with self datacdf = empcdf(y,y); % this gives the empirical quantile on each entry y % converts all values to uniform [0,1] - very useful Compare to theoretical model, GPD. (6.2.10) Fall 2016: LeBaron Fin285a: / 42

18 Example: fitgpareto.m Second, plot empirical and theoretical CDF s. % cdf from data datacdf = empcdf(tailret, tailret); % cdf from pareto gpdcdf = gpcdf(tailret,pparams(1),pparams(2),0); % convert to Prob(X>x X>u) datapgreater = 1-dataCDF; gpdpgreater = 1-gpdCDF; % mutiply by Prob(X>u) to get true tail, Prob(X>x) datapgreater = datapgreater * tailprob; gpdpgreater = gpdpgreater * tailprob; % remember to add threshold for return in plotting loglog(tailret+uthresh,datapgreater, *b ); hold on loglog(tailret+uthresh,gpdpgreater, *r ); Fall 2016: LeBaron Fin285a: / 42

19 Example: fitgpareto.m Third, get tail returns at probs. probs = [ ]; % get theoretical quantiles % remember to adjust for conditional tail prob adjustex = gpinv(1-probs/tailprob, pparams(1), pparams(2),0); % first add threshold adjustex = adjustex + uthresh; % flip sign and add mean % these are now back to regular returns adjustex = -1*adjustEx + mret; Let s go run this code now. Fall 2016: LeBaron Fin285a: / 42

20 Extreme value distributions Generalized Pareto distributions Tail shapes Using power laws with returns Time aggregation Chebyshev s inequality Tail shapes Summary Fall 2016: LeBaron Fin285a: / 42

21 Generalized Pareto again F u (x) = G ξ,β (x) = { 1 (1+ξ x β ) 1 ξ ξ 0 1 e (x β ) ξ = 0 Most cases for stocks are ξ 0. In this case, as x β, Pr(X u x X > u) = F u (x) 1 A 1 x α α = 1/ξ (6.2.11) Pr(X u > x X > u) = 1 F u (x) A 1 x α (6.2.12) Pr(X u > x X > u)pr(x > u) = Pr(X > x) (6.2.13) α is known as the tail index. Pr(X > x) Ax α (6.2.14) Fall 2016: LeBaron Fin285a: / 42

22 Generalized Pareto again Pr(X > x) Ax α α = 1/ξ For symmetric (around zero) returns we would also have Pr(R < x) A x α (6.2.15) What does this say? A bunch of interesting/important results. Fall 2016: LeBaron Fin285a: / 42

23 Tail facts Pr(R < x) A x α 1.Estimate Generalized Pareto parameters (and use α = 1/ξ) 2.Plotting/linearity: log(pr(r < x)) = log(a) αlog( x ) 3.Scaling: Pr(R < 2x) = A 2x α = A2 α x α = 2 α Pr(R < x) 4.Moments: (don t underestimate this fact) E(R m ) = m α Fall 2016: LeBaron Fin285a: / 42

24 Another tail fact: Student-t The student-t distribution with ν degrees of freedom has a power law tail with α = ν. This is also why the variance, E(R 2 ), doesn t exist for ν 2. Kurtosis, E(R 4 ), doesn t exist for ν 4. Fall 2016: LeBaron Fin285a: / 42

25 Another tail fact: Normal distributions What about a normal distribution? For the normal, α = All moments exist for a normal This corresponds to the special case of ξ = 0 for the Generalized Pareto If that parameter is near zero, then it is likely your data is normal As we know, most higher frequency financial data does not look normal Fall 2016: LeBaron Fin285a: / 42

26 Yet Another tail fact: Expected shortfall This only works out in the tail (small p), but it can be useful. ES(p) VaR(p) 1 1 ξ = 1 1 (1/α) (6.2.16) Fall 2016: LeBaron Fin285a: / 42

27 Dow return left tail (EV, α = 3.5) 10-3 Prob(X<ret) Data EV T(3) T(4) Left Tail Return Fall 2016: LeBaron Fin285a: / 42

28 Dow return left tail with normal Data EV T(3) T(4) Normal Prob(X<ret) Left Tail Return Fall 2016: LeBaron Fin285a: / 42

29 Extreme value distributions Generalized Pareto distributions Tail shapes Using power laws with returns Time aggregation Chebyshev s inequality Using power laws with returns Summary Fall 2016: LeBaron Fin285a: / 42

30 Estimate just α Estimate α Hill estimator (9.3.2) See also, LeBaron, Robust properties of stock return tails Picture/least squares (next slide) Big problem: determine tail region See (fig 9.3) Bias: extreme tail Variance: less extreme tail Fall 2016: LeBaron Fin285a: / 42

31 Least squares estimator Determine tail! Get x i values in tail log(pr(r < x)) = log(a) αlog( x ) Estimate (Pr(R < x i ), x i ) from data Take logs Estimate (A,α) using ordinary least squares regression Fall 2016: LeBaron Fin285a: / 42

32 Extreme value distributions Generalized Pareto distributions Tail shapes Using power laws with returns Time aggregation Chebyshev s inequality Time aggregation Summary Fall 2016: LeBaron Fin285a: / 42

33 Tail properties and sums X = X 1 +X 2 (6.2.17) Pr(X i < x) = A i x α (6.2.18) Fall 2016: LeBaron Fin285a: / 42

34 Tail properties and sums X = X 1 +X 2 (6.2.17) Pr(X i < x) = A i x α (6.2.18) Pr(X < x) = A x α (6.2.19) Fall 2016: LeBaron Fin285a: / 42

35 Tail properties and sums X = X 1 +X 2 (6.2.17) Pr(X i < x) = A i x α (6.2.18) See Thm 9.1, case 1 Pr(X < x) = A x α (6.2.19) Think about log returns and horizons r t (2) = r t (1)+r t 1 (1) Trick: estimate power laws with higher frequency data Fall 2016: LeBaron Fin285a: / 42

36 VaR and time scaling Thm 9.2 (page 178) VaR scales as T 1/α, T is time horizon If α = 4, then VaR scales as T 0.25 VaR 10 = VaR 1 For normal returns VaR scales with σ, the standard deviation This goes as Tσ 1 VaR 10 = VaR 1 Power law VaR increases more slowly Fall 2016: LeBaron Fin285a: / 42

37 Danielson table 9.1 VaR for starting value of 100. Tail index is about 4.3. VaR p-level 1% 0.1% 0.05% Extreme value (1 day) Extreme value (10 day) Normal (1 day) Normal (10 day) Note: EVT gives larger values at 1 day, but scales more slowly to 10 days < Square-root of two law is a conservative risk measure here. This is all pretty far out in the tail. Fall 2016: LeBaron Fin285a: / 42

38 Extreme value distributions Generalized Pareto distributions Tail shapes Using power laws with returns Time aggregation Chebyshev s inequality Chebyshev s inequality Summary Fall 2016: LeBaron Fin285a: / 42

39 Chebyshev s inequality Probability bound that holds for any distribution Very conservative risk measure Pr( X µ nσ) 1 n 2 (6.2.20) p = Pr(X µ nσ) 1 n 2 (6.2.21) n = 1 p (6.2.22) Software: ChebyshevComp.m Fall 2016: LeBaron Fin285a: / 42

40 Extreme value distributions Generalized Pareto distributions Tail shapes Using power laws with returns Time aggregation Chebyshev s inequality Summary Summary Fall 2016: LeBaron Fin285a: / 42

41 Extreme value useful features Can improve tail probability estimates Useful when samples are Small or, You want to go far out into the tail Caution: Tail index (α) can be difficult to estimate Can be combined with empirical bootstrap methods Use Generalized Pareto in tail and empirical density in middle of distribution Fall 2016: LeBaron Fin285a: / 42

42 Several methods Generalized Pareto Student-t Power-law Chebyshev (very conservative) Fall 2016: LeBaron Fin285a: / 42

43 Complicated?? This all seemed rather complicated There is a lot of fancy math behind this Why did we do this? Fall 2016: LeBaron Fin285a: / 42

44 Complicated?? This all seemed rather complicated There is a lot of fancy math behind this Why did we do this? Beyond fancy math, these distributions (power law, Pareto, extreme value) can be good approximations in the tail Probabilities for extreme risks Fall 2016: LeBaron Fin285a: / 42

45 Overview Extreme value distributions Generalized Pareto distributions Tail shapes Using power laws with returns Time aggregation Chebyshev s inequality Summary Fall 2016: LeBaron Fin285a: / 42

Financial Econometrics and Volatility Models Extreme Value Theory

Financial 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 information

Introduction to Algorithmic Trading Strategies Lecture 10

Introduction 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

Fin285a:Computer Simulations and Risk Assessment Section 2.3.2:Hypothesis testing, and Confidence Intervals

Fin285a:Computer Simulations and Risk Assessment Section 2.3.2:Hypothesis testing, and Confidence Intervals Fin285a:Computer Simulations and Risk Assessment Section 2.3.2:Hypothesis testing, and Confidence Intervals Overview Hypothesis testing terms Testing a die Testing issues Estimating means Confidence intervals

More information

Extreme Value Theory.

Extreme Value Theory. Bank of England Centre for Central Banking Studies CEMLA 2013 Extreme Value Theory. David G. Barr November 21, 2013 Any views expressed are those of the author and not necessarily those of the Bank of

More information

Overview of Extreme Value Theory. Dr. Sawsan Hilal space

Overview of Extreme Value Theory. Dr. Sawsan Hilal space Overview of Extreme Value Theory Dr. Sawsan Hilal space Maths Department - University of Bahrain space November 2010 Outline Part-1: Univariate Extremes Motivation Threshold Exceedances Part-2: Bivariate

More information

Zwiers FW and Kharin VV Changes in the extremes of the climate simulated by CCC GCM2 under CO 2 doubling. J. Climate 11:

Zwiers FW and Kharin VV Changes in the extremes of the climate simulated by CCC GCM2 under CO 2 doubling. J. Climate 11: Statistical Analysis of EXTREMES in GEOPHYSICS Zwiers FW and Kharin VV. 1998. Changes in the extremes of the climate simulated by CCC GCM2 under CO 2 doubling. J. Climate 11:2200 2222. http://www.ral.ucar.edu/staff/ericg/readinggroup.html

More information

Shape of the return probability density function and extreme value statistics

Shape of the return probability density function and extreme value statistics Shape of the return probability density function and extreme value statistics 13/09/03 Int. Workshop on Risk and Regulation, Budapest Overview I aim to elucidate a relation between one field of research

More information

Extreme value theory and high quantile convergence

Extreme value theory and high quantile convergence Journal of Operational Risk 51 57) Volume 1/Number 2, Summer 2006 Extreme value theory and high quantile convergence Mikhail Makarov EVMTech AG, Baarerstrasse 2, 6300 Zug, Switzerland In this paper we

More information

Math 576: Quantitative Risk Management

Math 576: Quantitative Risk Management Math 576: Quantitative Risk Management Haijun Li lih@math.wsu.edu Department of Mathematics Washington State University Week 11 Haijun Li Math 576: Quantitative Risk Management Week 11 1 / 21 Outline 1

More information

Distributions of Functions of Random Variables. 5.1 Functions of One Random Variable

Distributions of Functions of Random Variables. 5.1 Functions of One Random Variable Distributions of Functions of Random Variables 5.1 Functions of One Random Variable 5.2 Transformations of Two Random Variables 5.3 Several Random Variables 5.4 The Moment-Generating Function Technique

More information

Extreme Value Theory as a Theoretical Background for Power Law Behavior

Extreme Value Theory as a Theoretical Background for Power Law Behavior Extreme Value Theory as a Theoretical Background for Power Law Behavior Simone Alfarano 1 and Thomas Lux 2 1 Department of Economics, University of Kiel, alfarano@bwl.uni-kiel.de 2 Department of Economics,

More information

MFM Practitioner Module: Quantitiative Risk Management. John Dodson. October 14, 2015

MFM Practitioner Module: Quantitiative Risk Management. John Dodson. October 14, 2015 MFM Practitioner Module: Quantitiative Risk Management October 14, 2015 The n-block maxima 1 is a random variable defined as M n max (X 1,..., X n ) for i.i.d. random variables X i with distribution function

More information

Frequency Estimation of Rare Events by Adaptive Thresholding

Frequency Estimation of Rare Events by Adaptive Thresholding Frequency Estimation of Rare Events by Adaptive Thresholding J. R. M. Hosking IBM Research Division 2009 IBM Corporation Motivation IBM Research When managing IT systems, there is a need to identify transactions

More information

Wei-han Liu Department of Banking and Finance Tamkang University. R/Finance 2009 Conference 1

Wei-han Liu Department of Banking and Finance Tamkang University. R/Finance 2009 Conference 1 Detecting Structural Breaks in Tail Behavior -From the Perspective of Fitting the Generalized Pareto Distribution Wei-han Liu Department of Banking and Finance Tamkang University R/Finance 2009 Conference

More information

Overview of Extreme Value Analysis (EVA)

Overview of Extreme Value Analysis (EVA) Overview of Extreme Value Analysis (EVA) Brian Reich North Carolina State University July 26, 2016 Rossbypalooza Chicago, IL Brian Reich Overview of Extreme Value Analysis (EVA) 1 / 24 Importance of extremes

More information

Efficient Estimation of Distributional Tail Shape and the Extremal Index with Applications to Risk Management

Efficient Estimation of Distributional Tail Shape and the Extremal Index with Applications to Risk Management Journal of Mathematical Finance, 2016, 6, 626-659 http://www.scirp.org/journal/jmf ISSN Online: 2162-2442 ISSN Print: 2162-2434 Efficient Estimation of Distributional Tail Shape and the Extremal Index

More information

F X (x) = P [X x] = x f X (t)dt. 42 Lebesgue-a.e, to be exact 43 More specifically, if g = f Lebesgue-a.e., then g is also a pdf for X.

F X (x) = P [X x] = x f X (t)dt. 42 Lebesgue-a.e, to be exact 43 More specifically, if g = f Lebesgue-a.e., then g is also a pdf for X. 10.2 Properties of PDF and CDF for Continuous Random Variables 10.18. The pdf f X is determined only almost everywhere 42. That is, given a pdf f for a random variable X, if we construct a function g by

More information

Heteroskedasticity in Time Series

Heteroskedasticity in Time Series Heteroskedasticity in Time Series Figure: Time Series of Daily NYSE Returns. 206 / 285 Key Fact 1: Stock Returns are Approximately Serially Uncorrelated Figure: Correlogram of Daily Stock Market Returns.

More information

Sharp statistical tools Statistics for extremes

Sharp statistical tools Statistics for extremes Sharp statistical tools Statistics for extremes Georg Lindgren Lund University October 18, 2012 SARMA Background Motivation We want to predict outside the range of observations Sums, averages and proportions

More information

Quantitative Modeling of Operational Risk: Between g-and-h and EVT

Quantitative Modeling of Operational Risk: Between g-and-h and EVT : Between g-and-h and EVT Paul Embrechts Matthias Degen Dominik Lambrigger ETH Zurich (www.math.ethz.ch/ embrechts) Outline Basel II LDA g-and-h Aggregation Conclusion and References What is Basel II?

More information

Emma Simpson. 6 September 2013

Emma Simpson. 6 September 2013 6 September 2013 Test What is? Beijing during periods of low and high air pollution Air pollution is composed of sulphur oxides, nitrogen oxides, carbon monoxide and particulates. Particulates are small

More information

SYSM 6303: Quantitative Introduction to Risk and Uncertainty in Business Lecture 4: Fitting Data to Distributions

SYSM 6303: Quantitative Introduction to Risk and Uncertainty in Business Lecture 4: Fitting Data to Distributions SYSM 6303: Quantitative Introduction to Risk and Uncertainty in Business Lecture 4: Fitting Data to Distributions M. Vidyasagar Cecil & Ida Green Chair The University of Texas at Dallas Email: M.Vidyasagar@utdallas.edu

More information

M(t) = 1 t. (1 t), 6 M (0) = 20 P (95. X i 110) i=1

M(t) = 1 t. (1 t), 6 M (0) = 20 P (95. X i 110) i=1 Math 66/566 - Midterm Solutions NOTE: These solutions are for both the 66 and 566 exam. The problems are the same until questions and 5. 1. The moment generating function of a random variable X is M(t)

More information

Summary of basic probability theory Math 218, Mathematical Statistics D Joyce, Spring 2016

Summary of basic probability theory Math 218, Mathematical Statistics D Joyce, Spring 2016 8. For any two events E and F, P (E) = P (E F ) + P (E F c ). Summary of basic probability theory Math 218, Mathematical Statistics D Joyce, Spring 2016 Sample space. A sample space consists of a underlying

More information

Extreme Value Analysis and Spatial Extremes

Extreme 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 information

The Fundamentals of Heavy Tails Properties, Emergence, & Identification. Jayakrishnan Nair, Adam Wierman, Bert Zwart

The Fundamentals of Heavy Tails Properties, Emergence, & Identification. Jayakrishnan Nair, Adam Wierman, Bert Zwart The Fundamentals of Heavy Tails Properties, Emergence, & Identification Jayakrishnan Nair, Adam Wierman, Bert Zwart Why am I doing a tutorial on heavy tails? Because we re writing a book on the topic Why

More information

Volatility. Gerald P. Dwyer. February Clemson University

Volatility. Gerald P. Dwyer. February Clemson University Volatility Gerald P. Dwyer Clemson University February 2016 Outline 1 Volatility Characteristics of Time Series Heteroskedasticity Simpler Estimation Strategies Exponentially Weighted Moving Average Use

More information

Modelação de valores extremos e sua importância na

Modelação de valores extremos e sua importância na Modelação de valores extremos e sua importância na segurança e saúde Margarida Brito Departamento de Matemática FCUP (FCUP) Valores Extremos - DemSSO 1 / 12 Motivation Consider the following events Occurance

More information

Three hours. To be supplied by the Examinations Office: Mathematical Formula Tables and Statistical Tables THE UNIVERSITY OF MANCHESTER.

Three hours. To be supplied by the Examinations Office: Mathematical Formula Tables and Statistical Tables THE UNIVERSITY OF MANCHESTER. Three hours To be supplied by the Examinations Office: Mathematical Formula Tables and Statistical Tables THE UNIVERSITY OF MANCHESTER EXTREME VALUES AND FINANCIAL RISK Examiner: Answer QUESTION 1, QUESTION

More information

Extreme value modelling of rainfalls and

Extreme value modelling of rainfalls and Universite de Paris Sud Master s Thesis Extreme value modelling of rainfalls and flows Author: Fan JIA Supervisor: Pr Elisabeth Gassiat Dr Elena Di Bernardino Pr Michel Bera A thesis submitted in fulfilment

More information

Continuous Expectation and Variance, the Law of Large Numbers, and the Central Limit Theorem Spring 2014

Continuous Expectation and Variance, the Law of Large Numbers, and the Central Limit Theorem Spring 2014 Continuous Expectation and Variance, the Law of Large Numbers, and the Central Limit Theorem 18.5 Spring 214.5.4.3.2.1-4 -3-2 -1 1 2 3 4 January 1, 217 1 / 31 Expected value Expected value: measure of

More information

Analysis methods of heavy-tailed data

Analysis methods of heavy-tailed data Institute of Control Sciences Russian Academy of Sciences, Moscow, Russia February, 13-18, 2006, Bamberg, Germany June, 19-23, 2006, Brest, France May, 14-19, 2007, Trondheim, Norway PhD course Chapter

More information

Solutions of the Financial Risk Management Examination

Solutions of the Financial Risk Management Examination Solutions of the Financial Risk Management Examination Thierry Roncalli January 9 th 03 Remark The first five questions are corrected in TR-GDR and in the document of exercise solutions, which is available

More information

Practice Problems Section Problems

Practice Problems Section Problems Practice Problems Section 4-4-3 4-4 4-5 4-6 4-7 4-8 4-10 Supplemental Problems 4-1 to 4-9 4-13, 14, 15, 17, 19, 0 4-3, 34, 36, 38 4-47, 49, 5, 54, 55 4-59, 60, 63 4-66, 68, 69, 70, 74 4-79, 81, 84 4-85,

More information

12 Statistical Justifications; the Bias-Variance Decomposition

12 Statistical Justifications; the Bias-Variance Decomposition Statistical Justifications; the Bias-Variance Decomposition 65 12 Statistical Justifications; the Bias-Variance Decomposition STATISTICAL JUSTIFICATIONS FOR REGRESSION [So far, I ve talked about regression

More information

Reliable Inference in Conditions of Extreme Events. Adriana Cornea

Reliable Inference in Conditions of Extreme Events. Adriana Cornea Reliable Inference in Conditions of Extreme Events by Adriana Cornea University of Exeter Business School Department of Economics ExISta Early Career Event October 17, 2012 Outline of the talk Extreme

More information

1 Maintaining a Dictionary

1 Maintaining a Dictionary 15-451/651: Design & Analysis of Algorithms February 1, 2016 Lecture #7: Hashing last changed: January 29, 2016 Hashing is a great practical tool, with an interesting and subtle theory too. In addition

More information

Confidence intervals

Confidence intervals Confidence intervals We now want to take what we ve learned about sampling distributions and standard errors and construct confidence intervals. What are confidence intervals? Simply an interval for which

More information

Chapter 4. Probability Distributions Continuous

Chapter 4. Probability Distributions Continuous 1 Chapter 4 Probability Distributions Continuous Thus far, we have considered discrete pdfs (sometimes called probability mass functions) and have seen how that probability of X equaling a single number

More information

CS 5014: Research Methods in Computer Science. Bernoulli Distribution. Binomial Distribution. Poisson Distribution. Clifford A. Shaffer.

CS 5014: Research Methods in Computer Science. Bernoulli Distribution. Binomial Distribution. Poisson Distribution. Clifford A. Shaffer. Department of Computer Science Virginia Tech Blacksburg, Virginia Copyright c 2015 by Clifford A. Shaffer Computer Science Title page Computer Science Clifford A. Shaffer Fall 2015 Clifford A. Shaffer

More information

Exam C Solutions Spring 2005

Exam C Solutions Spring 2005 Exam C Solutions Spring 005 Question # The CDF is F( x) = 4 ( + x) Observation (x) F(x) compare to: Maximum difference 0. 0.58 0, 0. 0.58 0.7 0.880 0., 0.4 0.680 0.9 0.93 0.4, 0.6 0.53. 0.949 0.6, 0.8

More information

Modelling Operational Risk Using Bayesian Inference

Modelling Operational Risk Using Bayesian Inference Pavel V. Shevchenko Modelling Operational Risk Using Bayesian Inference 4y Springer 1 Operational Risk and Basel II 1 1.1 Introduction to Operational Risk 1 1.2 Defining Operational Risk 4 1.3 Basel II

More information

A primer on matrices

A primer on matrices A primer on matrices Stephen Boyd August 4, 2007 These notes describe the notation of matrices, the mechanics of matrix manipulation, and how to use matrices to formulate and solve sets of simultaneous

More information

Financial Econometrics and Quantitative Risk Managenent Return Properties

Financial 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 information

CHAPTER 1. Review of Algebra

CHAPTER 1. Review of Algebra CHAPTER 1 Review of Algebra Much of the material in this chapter is revision from GCSE maths (although some of the exercises are harder). Some of it particularly the work on logarithms may be new if you

More information

Question Points Score Total: 76

Question Points Score Total: 76 Math 447 Test 2 March 17, Spring 216 No books, no notes, only SOA-approved calculators. true/false or fill-in-the-blank question. You must show work, unless the question is a Name: Question Points Score

More information

Extreme Values and Their Applications in Finance

Extreme Values and Their Applications in Finance Extreme Values and Their Applications in Finance by Ruey S. Tsay Booth School of Business, The University of Chicago and Risk Management Institute, National University of Singapore Extreme price movements

More information

Chapter 5. Means and Variances

Chapter 5. Means and Variances 1 Chapter 5 Means and Variances Our discussion of probability has taken us from a simple classical view of counting successes relative to total outcomes and has brought us to the idea of a probability

More information

8 Laws of large numbers

8 Laws of large numbers 8 Laws of large numbers 8.1 Introduction We first start with the idea of standardizing a random variable. Let X be a random variable with mean µ and variance σ 2. Then Z = (X µ)/σ will be a random variable

More information

Lecture 12: Quality Control I: Control of Location

Lecture 12: Quality Control I: Control of Location Lecture 12: Quality Control I: Control of Location 10 October 2005 This lecture and the next will be about quality control methods. There are two reasons for this. First, it s intrinsically important for

More information

Note that we are looking at the true mean, μ, not y. The problem for us is that we need to find the endpoints of our interval (a, b).

Note that we are looking at the true mean, μ, not y. The problem for us is that we need to find the endpoints of our interval (a, b). Confidence Intervals 1) What are confidence intervals? Simply, an interval for which we have a certain confidence. For example, we are 90% certain that an interval contains the true value of something

More information

HIERARCHICAL MODELS IN EXTREME VALUE THEORY

HIERARCHICAL MODELS IN EXTREME VALUE THEORY HIERARCHICAL MODELS IN EXTREME VALUE THEORY Richard L. Smith Department of Statistics and Operations Research, University of North Carolina, Chapel Hill and Statistical and Applied Mathematical Sciences

More information

Most data analysis starts with some data set; we will call this data set P. It will be composed of a set of n

Most data analysis starts with some data set; we will call this data set P. It will be composed of a set of n 3 Convergence This topic will overview a variety of extremely powerful analysis results that span statistics, estimation theorem, and big data. It provides a framework to think about how to aggregate more

More information

The Growth of Functions. A Practical Introduction with as Little Theory as possible

The Growth of Functions. A Practical Introduction with as Little Theory as possible The Growth of Functions A Practical Introduction with as Little Theory as possible Complexity of Algorithms (1) Before we talk about the growth of functions and the concept of order, let s discuss why

More information

Financial Econometrics

Financial Econometrics Financial Econometrics Estimation and Inference Gerald P. Dwyer Trinity College, Dublin January 2013 Who am I? Visiting Professor and BB&T Scholar at Clemson University Federal Reserve Bank of Atlanta

More information

Bias Variance Trade-off

Bias Variance Trade-off Bias Variance Trade-off The mean squared error of an estimator MSE(ˆθ) = E([ˆθ θ] 2 ) Can be re-expressed MSE(ˆθ) = Var(ˆθ) + (B(ˆθ) 2 ) MSE = VAR + BIAS 2 Proof MSE(ˆθ) = E((ˆθ θ) 2 ) = E(([ˆθ E(ˆθ)]

More information

Extreme Value Theory and Applications

Extreme Value Theory and Applications Extreme Value Theory and Deauville - 04/10/2013 Extreme Value Theory and Introduction Asymptotic behavior of the Sum Extreme (from Latin exter, exterus, being on the outside) : Exceeding the ordinary,

More information

Markov Chains and Related Matters

Markov Chains and Related Matters Markov Chains and Related Matters 2 :9 3 4 : The four nodes are called states. The numbers on the arrows are called transition probabilities. For example if we are in state, there is a probability of going

More information

Generalized additive modelling of hydrological sample extremes

Generalized additive modelling of hydrological sample extremes Generalized additive modelling of hydrological sample extremes Valérie Chavez-Demoulin 1 Joint work with A.C. Davison (EPFL) and Marius Hofert (ETHZ) 1 Faculty of Business and Economics, University of

More information

This does not cover everything on the final. Look at the posted practice problems for other topics.

This does not cover everything on the final. Look at the posted practice problems for other topics. Class 7: Review Problems for Final Exam 8.5 Spring 7 This does not cover everything on the final. Look at the posted practice problems for other topics. To save time in class: set up, but do not carry

More information

A Conditional Approach to Modeling Multivariate Extremes

A Conditional Approach to Modeling Multivariate Extremes A Approach to ing Multivariate Extremes By Heffernan & Tawn Department of Statistics Purdue University s April 30, 2014 Outline s s Multivariate Extremes s A central aim of multivariate extremes is trying

More information

Regression Estimation Least Squares and Maximum Likelihood

Regression Estimation Least Squares and Maximum Likelihood Regression Estimation Least Squares and Maximum Likelihood Dr. Frank Wood Frank Wood, fwood@stat.columbia.edu Linear Regression Models Lecture 3, Slide 1 Least Squares Max(min)imization Function to minimize

More information

Model Fitting. Jean Yves Le Boudec

Model Fitting. Jean Yves Le Boudec Model Fitting Jean Yves Le Boudec 0 Contents 1. What is model fitting? 2. Linear Regression 3. Linear regression with norm minimization 4. Choosing a distribution 5. Heavy Tail 1 Virus Infection Data We

More information

{X i } realize. n i=1 X i. Note that again X is a random variable. If we are to

{X i } realize. n i=1 X i. Note that again X is a random variable. If we are to 3 Convergence This topic will overview a variety of extremely powerful analysis results that span statistics, estimation theorem, and big data. It provides a framework to think about how to aggregate more

More information

Using statistical methods to analyse environmental extremes.

Using statistical methods to analyse environmental extremes. Using statistical methods to analyse environmental extremes. Emma Eastoe Department of Mathematics and Statistics Lancaster University December 16, 2008 Focus of talk Discuss statistical models used to

More information

Bayesian Modelling of Extreme Rainfall Data

Bayesian Modelling of Extreme Rainfall Data Bayesian Modelling of Extreme Rainfall Data Elizabeth Smith A thesis submitted for the degree of Doctor of Philosophy at the University of Newcastle upon Tyne September 2005 UNIVERSITY OF NEWCASTLE Bayesian

More information

3.1 Inequalities - Graphing and Solving

3.1 Inequalities - Graphing and Solving 3.1 Inequalities - Graphing and Solving When we have an equation such as x = 4 we have a specific value for our variable. With inequalities we will give a range of values for our variable. To do this we

More information

Basic Probability Reference Sheet

Basic Probability Reference Sheet February 27, 2001 Basic Probability Reference Sheet 17.846, 2001 This is intended to be used in addition to, not as a substitute for, a textbook. X is a random variable. This means that X is a variable

More information

Change Point Analysis of Extreme Values

Change Point Analysis of Extreme Values Change Point Analysis of Extreme Values TIES 2008 p. 1/? Change Point Analysis of Extreme Values Goedele Dierckx Economische Hogeschool Sint Aloysius, Brussels, Belgium e-mail: goedele.dierckx@hubrussel.be

More information

An overview of applied econometrics

An overview of applied econometrics An overview of applied econometrics Jo Thori Lind September 4, 2011 1 Introduction This note is intended as a brief overview of what is necessary to read and understand journal articles with empirical

More information

Note that we are looking at the true mean, μ, not y. The problem for us is that we need to find the endpoints of our interval (a, b).

Note that we are looking at the true mean, μ, not y. The problem for us is that we need to find the endpoints of our interval (a, b). Confidence Intervals 1) What are confidence intervals? Simply, an interval for which we have a certain confidence. For example, we are 90% certain that an interval contains the true value of something

More information

Lecture 18 Miscellaneous Topics in Multiple Regression

Lecture 18 Miscellaneous Topics in Multiple Regression Lecture 18 Miscellaneous Topics in Multiple Regression STAT 512 Spring 2011 Background Reading KNNL: 8.1-8.5,10.1, 11, 12 18-1 Topic Overview Polynomial Models (8.1) Interaction Models (8.2) Qualitative

More information

Probability Methods in Civil Engineering Prof. Dr. Rajib Maity Department of Civil Engineering Indian Institute of Technology, Kharagpur

Probability Methods in Civil Engineering Prof. Dr. Rajib Maity Department of Civil Engineering Indian Institute of Technology, Kharagpur Probability Methods in Civil Engineering Prof. Dr. Rajib Maity Department of Civil Engineering Indian Institute of Technology, Kharagpur Lecture No. # 33 Probability Models using Gamma and Extreme Value

More information

Estimation of Quantiles

Estimation 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 information

Confidence intervals CE 311S

Confidence intervals CE 311S CE 311S PREVIEW OF STATISTICS The first part of the class was about probability. P(H) = 0.5 P(T) = 0.5 HTTHHTTTTHHTHTHH If we know how a random process works, what will we see in the field? Preview of

More information

18.175: Lecture 8 Weak laws and moment-generating/characteristic functions

18.175: Lecture 8 Weak laws and moment-generating/characteristic functions 18.175: Lecture 8 Weak laws and moment-generating/characteristic functions Scott Sheffield MIT 18.175 Lecture 8 1 Outline Moment generating functions Weak law of large numbers: Markov/Chebyshev approach

More information

Regression Estimation - Least Squares and Maximum Likelihood. Dr. Frank Wood

Regression Estimation - Least Squares and Maximum Likelihood. Dr. Frank Wood Regression Estimation - Least Squares and Maximum Likelihood Dr. Frank Wood Least Squares Max(min)imization Function to minimize w.r.t. β 0, β 1 Q = n (Y i (β 0 + β 1 X i )) 2 i=1 Minimize this by maximizing

More information

Hydrological extremes. Hydrology Flood Estimation Methods Autumn Semester

Hydrological extremes. Hydrology Flood Estimation Methods Autumn Semester Hydrological extremes droughts floods 1 Impacts of floods Affected people Deaths Events [Doocy et al., PLoS, 2013] Recent events in CH and Europe Sardinia, Italy, Nov. 2013 Central Europe, 2013 Genoa and

More information

Nonparametric estimation of extreme risks from heavy-tailed distributions

Nonparametric estimation of extreme risks from heavy-tailed distributions Nonparametric estimation of extreme risks from heavy-tailed distributions Laurent GARDES joint work with Jonathan EL METHNI & Stéphane GIRARD December 2013 1 Introduction to risk measures 2 Introduction

More information

Chapter 6 The Standard Deviation as a Ruler and the Normal Model

Chapter 6 The Standard Deviation as a Ruler and the Normal Model Chapter 6 The Standard Deviation as a Ruler and the Normal Model Overview Key Concepts Understand how adding (subtracting) a constant or multiplying (dividing) by a constant changes the center and/or spread

More information

APPLICATION OF EXTREMAL THEORY TO THE PRECIPITATION SERIES IN NORTHERN MORAVIA

APPLICATION OF EXTREMAL THEORY TO THE PRECIPITATION SERIES IN NORTHERN MORAVIA APPLICATION OF EXTREMAL THEORY TO THE PRECIPITATION SERIES IN NORTHERN MORAVIA DANIELA JARUŠKOVÁ Department of Mathematics, Czech Technical University, Prague; jarus@mat.fsv.cvut.cz 1. Introduction The

More information

High-frequency data modelling using Hawkes processes

High-frequency data modelling using Hawkes processes Valérie Chavez-Demoulin joint work with High-frequency A.C. Davison data modelling and using A.J. Hawkes McNeil processes(2005), J.A EVT2013 McGill 1 /(201 High-frequency data modelling using Hawkes processes

More information

Common ontinuous random variables

Common ontinuous random variables Common ontinuous random variables CE 311S Earlier, we saw a number of distribution families Binomial Negative binomial Hypergeometric Poisson These were useful because they represented common situations:

More information

Summary of Derivative Tests

Summary of Derivative Tests Summary of Derivative Tests Note that for all the tests given below it is assumed that the function f is continuous. Critical Numbers Definition. A critical number of a function f is a number c in the

More information

Asymptotic distribution of the sample average value-at-risk in the case of heavy-tailed returns

Asymptotic distribution of the sample average value-at-risk in the case of heavy-tailed returns Asymptotic distribution of the sample average value-at-risk in the case of heavy-tailed returns Stoyan V. Stoyanov Chief Financial Researcher, FinAnalytica Inc., Seattle, USA e-mail: stoyan.stoyanov@finanalytica.com

More information

Multivariate Distributions

Multivariate Distributions IEOR E4602: Quantitative Risk Management Spring 2016 c 2016 by Martin Haugh Multivariate Distributions We will study multivariate distributions in these notes, focusing 1 in particular on multivariate

More information

Probability. Table of contents

Probability. Table of contents Probability Table of contents 1. Important definitions 2. Distributions 3. Discrete distributions 4. Continuous distributions 5. The Normal distribution 6. Multivariate random variables 7. Other continuous

More information

Better Bootstrap Confidence Intervals

Better Bootstrap Confidence Intervals by Bradley Efron University of Washington, Department of Statistics April 12, 2012 An example Suppose we wish to make inference on some parameter θ T (F ) (e.g. θ = E F X ), based on data We might suppose

More information

High-frequency data modelling using Hawkes processes

High-frequency data modelling using Hawkes processes High-frequency data modelling using Hawkes processes Valérie Chavez-Demoulin 1 joint work J.A McGill 1 Faculty of Business and Economics, University of Lausanne, Switzerland Boulder, April 2016 Boulder,

More information

Mathematica Project 3

Mathematica Project 3 Mathematica Project 3 Name: Section: Date: On your class s Sakai site, your instructor has placed 5 Mathematica notebooks. Please use the following table to determine which file you should select based

More information

HEAVY-TRAFFIC EXTREME-VALUE LIMITS FOR QUEUES

HEAVY-TRAFFIC EXTREME-VALUE LIMITS FOR QUEUES HEAVY-TRAFFIC EXTREME-VALUE LIMITS FOR QUEUES by Peter W. Glynn Department of Operations Research Stanford University Stanford, CA 94305-4022 and Ward Whitt AT&T Bell Laboratories Murray Hill, NJ 07974-0636

More information

Tempered Stable and Pareto Distributions: Predictions Under Uncertainty

Tempered Stable and Pareto Distributions: Predictions Under Uncertainty Tempered Stable and Pareto Distributions: Predictions Under Uncertainty Robert L Wolpert & Kerrie L Mengersen Duke University & Queensland University of Technology Session 2A: Foundations I Thu 1:50 2:15pm

More information

Richard L. Smith Department of Statistics and Operations Research University of North Carolina Chapel Hill, NC

Richard L. Smith Department of Statistics and Operations Research University of North Carolina Chapel Hill, NC EXTREME VALUE THEORY Richard L. Smith Department of Statistics and Operations Research University of North Carolina Chapel Hill, NC 27599-3260 rls@email.unc.edu AMS Committee on Probability and Statistics

More information

3 Continuous Random Variables

3 Continuous Random Variables Jinguo Lian Math437 Notes January 15, 016 3 Continuous Random Variables Remember that discrete random variables can take only a countable number of possible values. On the other hand, a continuous random

More information

Random Variate Generation

Random Variate Generation CPSC 405 Random Variate Generation 2007W T1 Handout These notes present techniques to generate samples of desired probability distributions, and some fundamental results and techiques. Some of this material

More information

SECTION 7.4: PARTIAL FRACTIONS. These Examples deal with rational expressions in x, but the methods here extend to rational expressions in y, t, etc.

SECTION 7.4: PARTIAL FRACTIONS. These Examples deal with rational expressions in x, but the methods here extend to rational expressions in y, t, etc. SECTION 7.4: PARTIAL FRACTIONS (Section 7.4: Partial Fractions) 7.14 PART A: INTRO A, B, C, etc. represent unknown real constants. Assume that our polynomials have real coefficients. These Examples deal

More information

Probability inequalities 11

Probability inequalities 11 Paninski, Intro. Math. Stats., October 5, 2005 29 Probability inequalities 11 There is an adage in probability that says that behind every limit theorem lies a probability inequality (i.e., a bound on

More information

COSC 341 Human Computer Interaction. Dr. Bowen Hui University of British Columbia Okanagan

COSC 341 Human Computer Interaction. Dr. Bowen Hui University of British Columbia Okanagan COSC 341 Human Computer Interaction Dr. Bowen Hui University of British Columbia Okanagan 1 Last Topic Distribution of means When it is needed How to build one (from scratch) Determining the characteristics

More information

Overview. Confidence Intervals Sampling and Opinion Polls Error Correcting Codes Number of Pet Unicorns in Ireland

Overview. Confidence Intervals Sampling and Opinion Polls Error Correcting Codes Number of Pet Unicorns in Ireland Overview Confidence Intervals Sampling and Opinion Polls Error Correcting Codes Number of Pet Unicorns in Ireland Confidence Intervals When a random variable lies in an interval a X b with a specified

More information

Regression Analysis. y t = β 1 x t1 + β 2 x t2 + β k x tk + ϵ t, t = 1,..., T,

Regression Analysis. y t = β 1 x t1 + β 2 x t2 + β k x tk + ϵ t, t = 1,..., T, Regression Analysis The multiple linear regression model with k explanatory variables assumes that the tth observation of the dependent or endogenous variable y t is described by the linear relationship

More information