Modeling Ultra-High-Frequency Multivariate Financial Data by Monte Carlo Simulation Methods

Size: px
Start display at page:

Download "Modeling Ultra-High-Frequency Multivariate Financial Data by Monte Carlo Simulation Methods"

Transcription

1 Outline Modeling Ultra-High-Frequency Multivariate Financial Data by Monte Carlo Simulation Methods Ph.D. Student: Supervisor: Marco Minozzo Dipartimento di Scienze Economiche Università degli Studi di Verona Ph.D. Student Presentation Verona, May 2

2 Outline Outline Ultra-High-Frequency Data (UHF) 2 Marked Doubly Stochastic Poisson Processes (DSPP) Financial Models For Equally spaced data Marked Point Process (MPP) and UHF Data Literature review for MPP Marked DSPP A DSPP model with a common intensity 3 Modeling framework for logreturns Logreturn Models Simulation studies Brief Conclusion 4 Filtering by RJMCMC Markov Chain Monte Carlo Reversible jump MCMC

3 Outline Outline Ultra-High-Frequency Data (UHF) 2 Marked Doubly Stochastic Poisson Processes (DSPP) Financial Models For Equally spaced data Marked Point Process (MPP) and UHF Data Literature review for MPP Marked DSPP A DSPP model with a common intensity 3 Modeling framework for logreturns Logreturn Models Simulation studies Brief Conclusion 4 Filtering by RJMCMC Markov Chain Monte Carlo Reversible jump MCMC

4 Outline Outline Ultra-High-Frequency Data (UHF) 2 Marked Doubly Stochastic Poisson Processes (DSPP) Financial Models For Equally spaced data Marked Point Process (MPP) and UHF Data Literature review for MPP Marked DSPP A DSPP model with a common intensity 3 Modeling framework for logreturns Logreturn Models Simulation studies Brief Conclusion 4 Filtering by RJMCMC Markov Chain Monte Carlo Reversible jump MCMC

5 Outline Outline Ultra-High-Frequency Data (UHF) 2 Marked Doubly Stochastic Poisson Processes (DSPP) Financial Models For Equally spaced data Marked Point Process (MPP) and UHF Data Literature review for MPP Marked DSPP A DSPP model with a common intensity 3 Modeling framework for logreturns Logreturn Models Simulation studies Brief Conclusion 4 Filtering by RJMCMC Markov Chain Monte Carlo Reversible jump MCMC

6 .UHF 2.Model 3.Return 4.Filtering Motivation Interest in understanding financial markets, which are the source of UHF data. Most of the econometric literatures is dealing with regularly spaced data (5-min; 5-min; daily). Evidence of strong co-movements in individual trading rate. Correlation measures between returns has direct application in portfolio management. Methodology Our modeling framework for Multi-UHF data is based on Doubly Stochastic Poisson Processes, developing some ideas from Bauwens(26). Our model will allow to capture co-movements and common component through the implementation of RJMCMC algorithm.

7 .UHF 2.Model 3.Return 4.Filtering Motivation Interest in understanding financial markets, which are the source of UHF data. Most of the econometric literatures is dealing with regularly spaced data (5-min; 5-min; daily). Evidence of strong co-movements in individual trading rate. Correlation measures between returns has direct application in portfolio management. Methodology Our modeling framework for Multi-UHF data is based on Doubly Stochastic Poisson Processes, developing some ideas from Bauwens(26). Our model will allow to capture co-movements and common component through the implementation of RJMCMC algorithm.

8 Outline.UHF 2.Model 3.Return 4.Filtering Ultra-High-Frequency Data (UHF) 2 Marked Doubly Stochastic Poisson Processes (DSPP) Financial Models For Equally spaced data Marked Point Process (MPP) and UHF Data Literature review for MPP Marked DSPP A DSPP model with a common intensity 3 Modeling framework for logreturns Logreturn Models Simulation studies Brief Conclusion 4 Filtering by RJMCMC Markov Chain Monte Carlo Reversible jump MCMC

9 .UHF 2.Model 3.Return 4.Filtering UHF Data Definition (UHF) UHF data also known as tick-by-tick data: each tick is one logical unit of information, like a quote or a transaction price, time of event. Data Source: Borsa di Milano. 7 Italian banks: Banco Popolare (POP), Mediobanca (MED),Banca popolare di Milano (MIL), MPS Banca (MPS), Banca Intesa SanPaolo (ISP), UBI Banca (UBI), Unicredit Banca (UCD). Period: 5 business days from 27//28 to 4//28; market open from 9:5 to 7:25 (3s).

10 .UHF 2.Model 3.Return 4.Filtering The total number and the average transactions for each asset during 27//28 and 4//28 (5 business days). Asset Total number Frequency per of transactions business day Banco Popolare Medio Banca Banca popolare di Milano MPS Banca Intesa San Paolo Banca UBI Banca Unicredit Banca

11 .UHF 2.Model 3.Return 4.Filtering Characteristics of UHF Data Irregular time spacing Price Discreteness Negative first order autocorrelation of logreturns Empirical synchronized correlation (Epps effect) First 5 minute transactions Banco Popolare Medio Banca 8.6 price(euro) time(second)

12 .UHF 2.Model 3.Return 4.Filtering Negative First-Order Autocorrelation.5 Autocorrelation for Banco Popolare.5 Autocorrelation for Medio Banca time lag Autocorrelation for MPS Banca time lag.5 Autocorrelation for Banca Popolare di Milano time lag Autocorrelation for Banca di Intesa SanPaolo time lag.5 Autocorrelation for UBI Banca time lag Autocorrelation for Unicredit Banca time lag time lag

13 .UHF 2.Model 3.Return 4.Filtering Synchronization Methods Linear Interpolation and Previous-tick Interpolation Linear interpolation: to interpolate time between t j and t j + with the following form: Y (t + i t) = Y j + t + i t t j (Y j t j + t + Y j ) j where the j refer to the most recent time index of time t + i t. Previous-tick interpolation: take the most recent value of the raw data, in formula, Y (t + i t) = Y j where the j refer to the most recent time index of time t + i t.

14 .UHF 2.Model 3.Return 4.Filtering A graphic example of interpolation Graphic example of interpolation price time

15 .UHF 2.Model 3.Return 4.Filtering Previous-tick synchronization-logreturn aggregation r logp logp r2 logp2 r3 logp3 r4 logp4 r5 logp5 r6 logp6 logreturn aggregation r t = logp 3 logp =logp 3 logp 2 +logp 2 logp +logp logp = r 3 + r 2 + r

16 .UHF 2.Model 3.Return 4.Filtering Epps effect: Intesa Sanpaolo Banca with others CORRELATION ISP POP CORRELATION ISP MED x 4 CORRELATION ISP MPS x 4 CORRELATION ISP MIL x 4 CORRELATION ISP UBI x 4 CORRELATION ISP UCD x 4 x 4

17 .UHF 2.Model 3.Return 4.Filtering Linear dependence: Scatter plot of ISP-UCD.. ISP UCD(m) ISP UCD(4m) ISP UCD(m).5. ISP UCD(2m) ISP UCD(3m) ISP UCD(6m)

18 .UHF 2.Model 3.Return 4.Filtering Epps effect: Medio Banca with others CORRELATION MED POP CORRELATION MED MPS x 4 CORRELATION MED MIL x 4 CORRELATION MED ISP x 4 CORRELATION MED UBI x 4 CORRELATION MED UCD x x 4

19 .UHF 2.Model 3.Return 4.Filtering Epps effect: Banco Popolare with others CORRELATION POP MED CORRELATION POP MPS x 4 CORRELATION POP MIL x 4 CORRELATION POP ISP x 4 CORRELATION POP UBI x 4 CORRELATION POP UCD x x 4

20 .UHF 2.Model 3.Return 4.Filtering Epps effect: MPS Banca with others CORRELATION MPS POP CORRELATION MPS MED x 4 CORRELATION MPS MIL x 4 CORRELATION MPS ISP x 4 CORRELATION MPS UBI x 4 CORRELATION MPS UCD x 4 x 4

21 .UHF 2.Model 3.Return 4.Filtering Epps effect: Banca Popolare di Milano with others CORRELATION MIL POP CORRELATION MIL MED x 4 CORRELATION MIL MPS x 4 CORRELATION MIL ISP x 4 CORRELATION MIL UBI x 4 CORRELATION MIL UCD x 4 x 4

22 .UHF 2.Model 3.Return 4.Filtering Epps effect: UBI Banca with others CORRELATION UBI POP CORRELATION UBI MED x 4 CORRELATION UBI MPS x 4 CORRELATION UBI MIL x 4 CORRELATION UBI ISP x 4 CORRELATION UBI UCD x 4 x 4

23 .UHF 2.Model 3.Return 4.Filtering Epps effect: Unicredit Banca with others CORRELATION UCD POP CORRELATION UCD MED x 4 CORRELATION UCD MPS x 4 CORRELATION UCD MIL x 4 CORRELATION UCD ISP Time Scale(Seconds) x 4 CORRELATION UCD UBI x 4 x 4

24 .UHF 2.Model 3.Return 4.Filtering Epps effect Epps effect: the correlation of returns decrease dramatically as the time intervals enter the intra-hour level. It is extensively observed in the high frequency data. Epps effect correlation coefficient time scale

25 Outline.UHF 2.Model 3.Return 4.Filtering 2..ARCH 2.2.MPP 2.3.MPP 2.4.DSPP 2.5.Common Fact Ultra-High-Frequency Data (UHF) 2 Marked Doubly Stochastic Poisson Processes (DSPP) Financial Models For Equally spaced data Marked Point Process (MPP) and UHF Data Literature review for MPP Marked DSPP A DSPP model with a common intensity 3 Modeling framework for logreturns Logreturn Models Simulation studies Brief Conclusion 4 Filtering by RJMCMC Markov Chain Monte Carlo Reversible jump MCMC

26 Outline.UHF 2.Model 3.Return 4.Filtering 2..ARCH 2.2.MPP 2.3.MPP 2.4.DSPP 2.5.Common Fact Ultra-High-Frequency Data (UHF) 2 Marked Doubly Stochastic Poisson Processes (DSPP) Financial Models For Equally spaced data Marked Point Process (MPP) and UHF Data Literature review for MPP Marked DSPP A DSPP model with a common intensity 3 Modeling framework for logreturns Logreturn Models Simulation studies Brief Conclusion 4 Filtering by RJMCMC Markov Chain Monte Carlo Reversible jump MCMC

27 Literature review.uhf 2.Model 3.Return 4.Filtering 2..ARCH 2.2.MPP 2.3.MPP 2.4.DSPP 2.5.Common Fact Standard Econometric Models Engle(982): ARCH Model Bollerslev(986): GARCH

28 Outline.UHF 2.Model 3.Return 4.Filtering 2..ARCH 2.2.MPP 2.3.MPP 2.4.DSPP 2.5.Common Fact Ultra-High-Frequency Data (UHF) 2 Marked Doubly Stochastic Poisson Processes (DSPP) Financial Models For Equally spaced data Marked Point Process (MPP) and UHF Data Literature review for MPP Marked DSPP A DSPP model with a common intensity 3 Modeling framework for logreturns Logreturn Models Simulation studies Brief Conclusion 4 Filtering by RJMCMC Markov Chain Monte Carlo Reversible jump MCMC

29 .UHF 2.Model 3.Return 4.Filtering 2..ARCH 2.2.MPP 2.3.MPP 2.4.DSPP 2.5.Common Fact Marked Point Processes (MPP) and UHF Data T = T T 2 Z 2 Z 3 T 3 T 4 Marked Point Process A marked point process (MPP) is a sequence of pairs Z Z 4 φ = (T i, Z i ) i N. Time Ultra-High-Frequency (UHF) Data With ultra-high-frequency (UHF) data, all trades (or quotes) of a financial asset are recorded (time of event, (bid-ask) price, volume, etc.). In this case, T i and Z i, i =, 2,..., could be the time and the size of the ith logprice change.

30 .UHF 2.Model 3.Return 4.Filtering 2..ARCH 2.2.MPP 2.3.MPP 2.4.DSPP 2.5.Common Fact Marked Point Processes (MPP) and UHF Data T = T T 2 Z 2 Z 3 T 3 T 4 Marked Point Process A marked point process (MPP) is a sequence of pairs Z Z 4 φ = (T i, Z i ) i N. Time Ultra-High-Frequency (UHF) Data With ultra-high-frequency (UHF) data, all trades (or quotes) of a financial asset are recorded (time of event, (bid-ask) price, volume, etc.). In this case, T i and Z i, i =, 2,..., could be the time and the size of the ith logprice change.

31 Outline.UHF 2.Model 3.Return 4.Filtering 2..ARCH 2.2.MPP 2.3.MPP 2.4.DSPP 2.5.Common Fact Ultra-High-Frequency Data (UHF) 2 Marked Doubly Stochastic Poisson Processes (DSPP) Financial Models For Equally spaced data Marked Point Process (MPP) and UHF Data Literature review for MPP Marked DSPP A DSPP model with a common intensity 3 Modeling framework for logreturns Logreturn Models Simulation studies Brief Conclusion 4 Filtering by RJMCMC Markov Chain Monte Carlo Reversible jump MCMC

32 .UHF 2.Model 3.Return 4.Filtering Literature review for MPP 2..ARCH 2.2.MPP 2.3.MPP 2.4.DSPP 2.5.Common Fact Autoregressive Conditional Duration (ACD) Models Engle and Russell (998): Autoregressive Conditional Duration: A New Model for Irregularly Spaced Transaction Data ; Zhang, Russell and Tsay (2): A Nonlinear Autoregressive Conditional Duration Model with Applications to Financial Transaction Data ; Bauwens and Giot (2): Asymmetric ACD Models: Introducing Price Information in ACD Models.

33 .UHF 2.Model 3.Return 4.Filtering Literature Review for MPP 2..ARCH 2.2.MPP 2.3.MPP 2.4.DSPP 2.5.Common Fact Marked Doubly Stochastic Poisson Processes (DSPP) Rydberg and Shephard (2): A modelling framework for the prices and times of trades made on the New York stock exchange. ; Frey (2): Risk-minimization with incomplete information in a model for high-frequency data ; Frey and Runggaldier (2): A nonlinear filtering approach to volatility estimaiton with a view towards high frequency data ; Centanni and Minozzo (26): Estimaiton and filtering by reversible jump MCMC for a doubly stochastic Poisson model for ultra-high-frequency data ;

34 .UHF 2.Model 3.Return 4.Filtering Literature Review for MPP 2..ARCH 2.2.MPP 2.3.MPP 2.4.DSPP 2.5.Common Fact Multivarite Point Process Bowsher(22): Modelling Security Markets in Continuous Time: Intensity based, Multivariate Point Process Models ; Hall and Hautsch(24): A Continuous-Time Measurement of the Buy-Sell Pressure in a Limite Order Book Market ; Hall and Hautsch(26): Order Aggressiveness and Order Book Dynamics ; Hautsch(25): The Latent Factor VAR Model: Testing for a Common Component in the Intraday Trading Process ; Bauwens and Hautsch(26): Stochastic Conditional Intensity Process.

35 Outline.UHF 2.Model 3.Return 4.Filtering 2..ARCH 2.2.MPP 2.3.MPP 2.4.DSPP 2.5.Common Fact Ultra-High-Frequency Data (UHF) 2 Marked Doubly Stochastic Poisson Processes (DSPP) Financial Models For Equally spaced data Marked Point Process (MPP) and UHF Data Literature review for MPP Marked DSPP A DSPP model with a common intensity 3 Modeling framework for logreturns Logreturn Models Simulation studies Brief Conclusion 4 Filtering by RJMCMC Markov Chain Monte Carlo Reversible jump MCMC

36 .UHF 2.Model 3.Return 4.Filtering 2..ARCH 2.2.MPP 2.3.MPP 2.4.DSPP 2.5.Common Fact Marked Doubly Stochastic Poisson Processes (DSPP) Definition (Last and Brandt, 995) Given a probability space (Ω, F, P) and a filtration {F t } t R +, an MPP φ adapted to the filtration is a DSPP if there exists a F -measurable random measure ν on R + R such that ( P µ ( (s, t] A ) ) (υ ( (s, t] A )) k = k F s = k! ( ) e υ (s,t] A, almost surely, for every A B(R), where µ denotes the counting measure associated to the MPP φ, that is, µ ( ω, (, t] A ) N t = i= {Zi A}.

37 .UHF 2.Model 3.Return 4.Filtering 2..ARCH 2.2.MPP 2.3.MPP 2.4.DSPP 2.5.Common Fact Marked Doubly Stochastic Poisson Processes (DSPP) Intensity of a Marked DSPP DSPPs are point processes in which the number of jumps N t N s in any time interval (s, t] is Poisson distributed, given another positive stochastic process λ, called intensity P ( N t N s = k λ ) = ( t s λ udu k! ) k ( t ) exp λ u du. s Moreover, given λ, the number of jumps in disjoint time intervals are independent. Marked DSPP With Intensity Driven by MPP We assume the intensity λ depends itself on another MPP, that is, λ is of the form λ t = h(t, ψ t), where ψt is the restriction of another MPP ψ = (τ j, X j ) j N to [, t].

38 .UHF 2.Model 3.Return 4.Filtering 2..ARCH 2.2.MPP 2.3.MPP 2.4.DSPP 2.5.Common Fact Marked Doubly Stochastic Poisson Processes (DSPP) Intensity of a Marked DSPP DSPPs are point processes in which the number of jumps N t N s in any time interval (s, t] is Poisson distributed, given another positive stochastic process λ, called intensity P ( N t N s = k λ ) = ( t s λ udu k! ) k ( t ) exp λ u du. s Moreover, given λ, the number of jumps in disjoint time intervals are independent. Marked DSPP With Intensity Driven by MPP We assume the intensity λ depends itself on another MPP, that is, λ is of the form λ t = h(t, ψ t), where ψt is the restriction of another MPP ψ = (τ j, X j ) j N to [, t].

39 .UHF 2.Model 3.Return 4.Filtering 2..ARCH 2.2.MPP 2.3.MPP 2.4.DSPP 2.5.Common Fact Marked DSPP With Shot Noise Intensity (Basic Model) Stochastic Differential Equation Stochastic noise intensity is assumed as the solution of the following Stochastic Differential Equation: dλ t = kλ t dt + dj t where k >,,J = N t j= X j, and N t = #j : τ j t. Solution of SDE N t λ t = λ exp( kt) + X j exp( k(t τ j )), t j= where (τ j ) j=,2,... is a Poisson process with constant parameter ν, and X j are exponentially distributed with parameter γ.

40 .UHF 2.Model 3.Return 4.Filtering 2..ARCH 2.2.MPP 2.3.MPP 2.4.DSPP 2.5.Common Fact Marked DSPP With Shot Noise Intensity (Basic Model) Stochastic Differential Equation Stochastic noise intensity is assumed as the solution of the following Stochastic Differential Equation: dλ t = kλ t dt + dj t where k >,,J = N t j= X j, and N t = #j : τ j t. Solution of SDE N t λ t = λ exp( kt) + X j exp( k(t τ j )), t j= where (τ j ) j=,2,... is a Poisson process with constant parameter ν, and X j are exponentially distributed with parameter γ.

41 .UHF 2.Model 3.Return 4.Filtering 2..ARCH 2.2.MPP 2.3.MPP 2.4.DSPP 2.5.Common Fact Simulated Shot Noise Intensity with parameter ν =., k =., γ =.2 35 Intensity 3 25 Lambda Time

42 Outline.UHF 2.Model 3.Return 4.Filtering 2..ARCH 2.2.MPP 2.3.MPP 2.4.DSPP 2.5.Common Fact Ultra-High-Frequency Data (UHF) 2 Marked Doubly Stochastic Poisson Processes (DSPP) Financial Models For Equally spaced data Marked Point Process (MPP) and UHF Data Literature review for MPP Marked DSPP A DSPP model with a common intensity 3 Modeling framework for logreturns Logreturn Models Simulation studies Brief Conclusion 4 Filtering by RJMCMC Markov Chain Monte Carlo Reversible jump MCMC

43 .UHF 2.Model 3.Return 4.Filtering Common Factor Intensity Model 2..ARCH 2.2.MPP 2.3.MPP 2.4.DSPP 2.5.Common Fact Model Description λ s t = λ s t + α s λ t s =, 2,...,m. where λ s t denotes s-specific individual intensity and λ t denotes a common component. The coefficient s is a scaling parameter driving the common intensity impact on the s-individual intensity. Bivariate case λ t λ 2 t = λ t + α λ t = λ 2 t + α 2 λ t

44 .UHF 2.Model 3.Return 4.Filtering 2..ARCH 2.2.MPP 2.3.MPP 2.4.DSPP 2.5.Common Fact Simulated common intensity with parameter ν =.,k =.,γ =.2,ν =.5,k =.5,γ =,ν 2 =,k 2 =.4,γ 2 =. 3 Lambda Lambda Time Lambda Lambda2 Lambda 5 Lambda Time.5 Transaction time Time.5 Transaction time Time Time

45 Outline.UHF 2.Model 3.Return 4.Filtering 3..Return 3.2.ρ 3.3.Summary Ultra-High-Frequency Data (UHF) 2 Marked Doubly Stochastic Poisson Processes (DSPP) Financial Models For Equally spaced data Marked Point Process (MPP) and UHF Data Literature review for MPP Marked DSPP A DSPP model with a common intensity 3 Modeling framework for logreturns Logreturn Models Simulation studies Brief Conclusion 4 Filtering by RJMCMC Markov Chain Monte Carlo Reversible jump MCMC

46 Outline.UHF 2.Model 3.Return 4.Filtering 3..Return 3.2.ρ 3.3.Summary Ultra-High-Frequency Data (UHF) 2 Marked Doubly Stochastic Poisson Processes (DSPP) Financial Models For Equally spaced data Marked Point Process (MPP) and UHF Data Literature review for MPP Marked DSPP A DSPP model with a common intensity 3 Modeling framework for logreturns Logreturn Models Simulation studies Brief Conclusion 4 Filtering by RJMCMC Markov Chain Monte Carlo Reversible jump MCMC

47 .UHF 2.Model 3.Return 4.Filtering Modelling framework for logreturns 3..Return 3.2.ρ 3.3.Summary Independent Wiener Process R Ti = µ + ξ Ti, ξ Ti N (( ) ( )) σ, (T i T i ) σ 22 Concurrently correlated Wiener Process R Ti = µ + ξ Ti, (( ) ( )) σ σ 2 ξ t N, (T i T i ) σ 2 σ 22 where σ 2 = σ 2.

48 .UHF 2.Model 3.Return 4.Filtering Modelling framework for logreturns 3..Return 3.2.ρ 3.3.Summary Independent Wiener Process R Ti = µ + ξ Ti, ξ Ti N (( ) ( )) σ, (T i T i ) σ 22 Concurrently correlated Wiener Process R Ti = µ + ξ Ti, (( ) ( )) σ σ 2 ξ t N, (T i T i ) σ 2 σ 22 where σ 2 = σ 2.

49 .UHF 2.Model 3.Return 4.Filtering Modelling framework for logreturns 3..Return 3.2.ρ 3.3.Summary Bi-AR() without cross-correlation (( ) ( )) σ σ 2 R Ti = µ + βr Ti + ξ Ti, ξ Ti N, σ 2 σ 22 where β = ( β ) β 22 Bi-AR() R Ti = µ + βr Ti + ξ Ti, (( ) ( )) σ σ 2 ξ T i N, σ 2 σ 22 where β ij, i, j =, 2.

50 .UHF 2.Model 3.Return 4.Filtering Modelling framework for logreturns 3..Return 3.2.ρ 3.3.Summary Bi-AR() without cross-correlation (( ) ( )) σ σ 2 R Ti = µ + βr Ti + ξ Ti, ξ Ti N, σ 2 σ 22 where β = ( β ) β 22 Bi-AR() R Ti = µ + βr Ti + ξ Ti, (( ) ( )) σ σ 2 ξ T i N, σ 2 σ 22 where β ij, i, j =, 2.

51 Outline.UHF 2.Model 3.Return 4.Filtering 3..Return 3.2.ρ 3.3.Summary Ultra-High-Frequency Data (UHF) 2 Marked Doubly Stochastic Poisson Processes (DSPP) Financial Models For Equally spaced data Marked Point Process (MPP) and UHF Data Literature review for MPP Marked DSPP A DSPP model with a common intensity 3 Modeling framework for logreturns Logreturn Models Simulation studies Brief Conclusion 4 Filtering by RJMCMC Markov Chain Monte Carlo Reversible jump MCMC

52 Pooled time.uhf 2.Model 3.Return 4.Filtering 3..Return 3.2.ρ 3.3.Summary t t2 t3 t4 t5 t t2 t3 T T2 T3 T4 T5 T6 T7 T8 R R2 R3 R4 R5 R6 R7 R8 R R2 R3 R4 R5 R6 R7 R8

53 µ =.UHF 2.Model 3.Return 4.Filtering 3..Return 3.2.ρ 3.3.Summary i)independent Wiener Process with σ = σ 22 =., σ 2 = σ 2 =.3 Independent time scale

54 µ =.UHF 2.Model 3.Return 4.Filtering 3..Return 3.2.ρ 3.3.Summary ii) Concurrently correlated Wiener Process with σ =., σ 22 =.2, σ 2 = σ 2 =. Temporally Correlated Temporally Correlated time scale Temporally Correlated time scale 5 5 time scale

55 µ =.UHF 2.Model 3.Return 4.Filtering 3..Return 3.2.ρ 3.3.Summary iii) Bi-AR() without cross correlation: σ =., σ 22 =.2, σ 2 = σ 2 =., β =., β 22 =.2, β 2 = β 2 = VAR() without cross correlation time scale VAR() without cross correlation VAR() without cross correlation time scale 5 5 time scale

56 µ =.UHF 2.Model 3.Return 4.Filtering 3..Return 3.2.ρ 3.3.Summary iv) Bi-AR(): σ =., σ 22 =.2, σ 2 = σ 2 =., β =.2, β 22 =., β 2 =.3, β 2 =.6 VAR() time scale

57 Outline.UHF 2.Model 3.Return 4.Filtering 3..Return 3.2.ρ 3.3.Summary Ultra-High-Frequency Data (UHF) 2 Marked Doubly Stochastic Poisson Processes (DSPP) Financial Models For Equally spaced data Marked Point Process (MPP) and UHF Data Literature review for MPP Marked DSPP A DSPP model with a common intensity 3 Modeling framework for logreturns Logreturn Models Simulation studies Brief Conclusion 4 Filtering by RJMCMC Markov Chain Monte Carlo Reversible jump MCMC

58 Brief Conclusion.UHF 2.Model 3.Return 4.Filtering 3..Return 3.2.ρ 3.3.Summary Common factor model of intensity has no significant impact on Epps effect, in general on empirical synchronized correlation function; To obtain the empirical correlation, we need not only concurrent correlation but also cross-serial correlation; Epps effect caused by asynchronization. logp r2 r logp logp2 r3 logp3 r4 logp4 r5 logp5 r6 logp6 logreturn aggregation r t = logp 3 logp =logp 3 logp 2 +logp 2 logp +logp logp = r 3 + r 2 + r

59 Brief Conclusion.UHF 2.Model 3.Return 4.Filtering 3..Return 3.2.ρ 3.3.Summary Common factor model of intensity has no significant impact on Epps effect, in general on empirical synchronized correlation function; To obtain the empirical correlation, we need not only concurrent correlation but also cross-serial correlation; Epps effect caused by asynchronization. logp r2 r logp logp2 r3 logp3 r4 logp4 r5 logp5 r6 logp6 logreturn aggregation r t = logp 3 logp =logp 3 logp 2 +logp 2 logp +logp logp = r 3 + r 2 + r

60 .UHF 2.Model 3.Return 4.Filtering Cross-Autocorrelation 3..Return 3.2.ρ 3.3.Summary logp logp4 r logp r2 logp2 r3 r4 logp3 r5 logp5 r6 logp6 News logp logp logp2 logp3 News

61 Outline.UHF 2.Model 3.Return 4.Filtering 4..MCMC 4.2.RJMCMC Ultra-High-Frequency Data (UHF) 2 Marked Doubly Stochastic Poisson Processes (DSPP) Financial Models For Equally spaced data Marked Point Process (MPP) and UHF Data Literature review for MPP Marked DSPP A DSPP model with a common intensity 3 Modeling framework for logreturns Logreturn Models Simulation studies Brief Conclusion 4 Filtering by RJMCMC Markov Chain Monte Carlo Reversible jump MCMC

62 Outline.UHF 2.Model 3.Return 4.Filtering 4..MCMC 4.2.RJMCMC Ultra-High-Frequency Data (UHF) 2 Marked Doubly Stochastic Poisson Processes (DSPP) Financial Models For Equally spaced data Marked Point Process (MPP) and UHF Data Literature review for MPP Marked DSPP A DSPP model with a common intensity 3 Modeling framework for logreturns Logreturn Models Simulation studies Brief Conclusion 4 Filtering by RJMCMC Markov Chain Monte Carlo Reversible jump MCMC

63 .UHF 2.Model 3.Return 4.Filtering Markov Chain Monte Carlo(MCMC) 4..MCMC 4.2.RJMCMC Introduction MCMC methods are a class of algorithms for sampling from probability distribution based on constructing a Markov chain that has the target distribution (π) as its stationary distribution. In order for π to be the stationary distribution of the chain, the transition probabilities must satisfy the detailed balance condition π i P ij = π j P ji MCMC Two important algorithm to construct Markov Chain: Metropolis-Hasting algorithm; Gibbs sampling (which is the special case of M-H algorithm).

64 .UHF 2.Model 3.Return 4.Filtering Markov Chain Monte Carlo(MCMC) 4..MCMC 4.2.RJMCMC Introduction MCMC methods are a class of algorithms for sampling from probability distribution based on constructing a Markov chain that has the target distribution (π) as its stationary distribution. In order for π to be the stationary distribution of the chain, the transition probabilities must satisfy the detailed balance condition π i P ij = π j P ji MCMC Two important algorithm to construct Markov Chain: Metropolis-Hasting algorithm; Gibbs sampling (which is the special case of M-H algorithm).

65 .UHF 2.Model 3.Return 4.Filtering Metropolis-Hasting Algorithm 4..MCMC 4.2.RJMCMC Suppose we have a Markov Chain in state x. We want to simulate a draw from the transition kernel p(x, y) with stationary distribution π, but we do not know the form of p(x, y). We do know how to compute a function proportional to π, f (x) = kπ(x). Assume that we can draw y from an arbitrary distribution q(x, y). Consider using this q as a transition kernel. If we construct α(x, y) such that π(x)q(x, y)α(x, y) = π(y)q(y, x)α(y, x) then we will have a reversible transition kernel with stationary distribution π. We can take: α(x, y) = min{, π(y)q(y, x) π(x)q(x, y) } Although we do not know π(x), we do know f (x) = kπ(x), so we can calculate α(x, y).

66 .UHF 2.Model 3.Return 4.Filtering Metropolis-Hasting Algorithm 4..MCMC 4.2.RJMCMC In summary, the Metropolis-Hasting algorithm proceeds as follows. Given x j, to move to x j+ we:. Generate a draw y from q(x j,.) 2. Calculate α(x j, y) 3. Generate u U[, ] 4. if α(x j, y) > u, then x j+ = y. Otherwise x j+ = x j. After a large number of iterations, the marginal distribution of x j will converge to π. Example Suppose we have three intensity processes, λ = ν x, λ 2 = ν 2 x 2, λ = ν x, where x i Gamma(α i, β i ), and ν i, α i and β i are parameters, i =,, 2.

67 .UHF 2.Model 3.Return 4.Filtering Metropolis-Hasting Algorithm 4..MCMC 4.2.RJMCMC In summary, the Metropolis-Hasting algorithm proceeds as follows. Given x j, to move to x j+ we:. Generate a draw y from q(x j,.) 2. Calculate α(x j, y) 3. Generate u U[, ] 4. if α(x j, y) > u, then x j+ = y. Otherwise x j+ = x j. After a large number of iterations, the marginal distribution of x j will converge to π. Example Suppose we have three intensity processes, λ = ν x, λ 2 = ν 2 x 2, λ = ν x, where x i Gamma(α i, β i ), and ν i, α i and β i are parameters, i =,, 2.

68 .UHF 2.Model 3.Return 4.Filtering 4..MCMC 4.2.RJMCMC 4 Lambda=v*x 4 Lambda2=v2*x2 Lambda 3 2 Lambda Time Time Lambda=v*x Lambda Lambda=Lambda+a*Lambda Time Lambda2=Lambda2+a2*Lambda 4 2 Return Return Time Time

69 .UHF 2.Model 3.Return 4.Filtering 4..MCMC 4.2.RJMCMC Run of Metropolis-Hasting algorithm for iterations with 5 burn-in period. Parameter ν =,ν = 2, ν 2 =, α = 3, β =, α = 2, β = 2, α 2 = 5, β 2 =.5,a =, a 2 =.5. 8 Histogram of X x = Histogram of X x = Histogram of X x 2 =

70 Outline.UHF 2.Model 3.Return 4.Filtering 4..MCMC 4.2.RJMCMC Ultra-High-Frequency Data (UHF) 2 Marked Doubly Stochastic Poisson Processes (DSPP) Financial Models For Equally spaced data Marked Point Process (MPP) and UHF Data Literature review for MPP Marked DSPP A DSPP model with a common intensity 3 Modeling framework for logreturns Logreturn Models Simulation studies Brief Conclusion 4 Filtering by RJMCMC Markov Chain Monte Carlo Reversible jump MCMC

71 .UHF 2.Model 3.Return 4.Filtering Reversible jump MCMC 4..MCMC 4.2.RJMCMC Introduction RJMCMC algorithm is a generalization of the Metropolis-Hasting algorithm that supplies a sample from a target distribution π on the spaces of the form C = n= C n, where C n = {n} R kn, k n N, in which subspaces of different dimention k n are combined. RJMCMC We define r transition moves to choose the next state of the chain, where some of the moves involve vectors of different dimensions. Thus, RJMCMC algorithm update proceeds with the following steps: i) choose one of the moves; ii) propose a new state; iii) accept the new state with a certain probability A such that the Markov chain converges.

72 .UHF 2.Model 3.Return 4.Filtering Reversible jump MCMC 4..MCMC 4.2.RJMCMC Introduction RJMCMC algorithm is a generalization of the Metropolis-Hasting algorithm that supplies a sample from a target distribution π on the spaces of the form C = n= C n, where C n = {n} R kn, k n N, in which subspaces of different dimention k n are combined. RJMCMC We define r transition moves to choose the next state of the chain, where some of the moves involve vectors of different dimensions. Thus, RJMCMC algorithm update proceeds with the following steps: i) choose one of the moves; ii) propose a new state; iii) accept the new state with a certain probability A such that the Markov chain converges.

73 o t x x 5 3 x 4 =x* o τ 4 =τ* x4 x 4 t t t λ x 4 o t t.uhf 2.Model 3.Return 4.Filtering Move types of RJMCMC 4..MCMC 4.2.RJMCMC λ t = λ exp( kt) + N t j= X j exp( k(t τ j )) birth x x 2 height position current τ τ 2 τ 3 τ 5 ¹ x x 2 x 3 x 4 x 4 τ τ 2 τ 3 τ 4 x x 2 x 3 x 4 ¹ λ x x 2 x 3 τ τ 2 τ 2 τ 3 τ 4 τ τ 2 τ 3 τ 4 ¹ x x 2 x 3 death x 2 x 3 x τ τ 2 τ 3 τ 4 τ τ 2 τ 3 τ 4 starting value

74 .UHF 2.Model 3.Return 4.Filtering 4..MCMC 4.2.RJMCMC Component (s).3 (h) (p).5 (b) (d) (s).3 (h).3.3 (p).5.5 (b) (d) (s).3.3 (h).3 2 (p).5.5 (b) (d)

75 .UHF 2.Model 3.Return 4.Filtering 4..MCMC 4.2.RJMCMC Bivariate DSPP model filtering by RJMCMC Filtering of the Intensity by RJMCMC,T=,run for iteration with 5 burn in a= a2= 2 True(simulated)trajectories of the joint intensity Filtering expectation of the joint intensity 5 Lambda () Time 25 2 True(simulated)trajectories of the joint intensity Filtering expectation of the joint intensity Lambda (2) Time

76 .UHF 2.Model 3.Return 4.Filtering 4..MCMC 4.2.RJMCMC Bivariate DSPP model filtering by RJMCMC Filtering of the Intensity by RJMCMC,T=,run for iteration with 5 burn in a= a2=. True(simulated)trajectories of the common intensity Filtering expectation of the common intensity 8 Lambda Lambda Time True(simulated)trajectories of the first intensity Filtering expectation of the first intensity Lambda True(simulated)trajectories of the second intensity Filtering expectation of the second intensity Time Time

77 5.Conclusions Conclusion and Further research Bibliography A new model for multivariate DSPP based on a common component for the intensities. The intensities of DSPPs are not observable; RJMCMC methods are applied for filtering. VAR() provides correlations between two assets close to empirical ones (EPPs effect). Further research: Parameter estimation by Monte Carlo Expection-Maximization (MCEM) methods based on the RJMCMC algorithm.

78 Bibliography 5.Conclusions Bibliography L. Bauwens and N. Hautsch (26). Stochastic conditional intensity Process. Journal of Financial Econometrics, 4, S. Centanni and M. Minozzo (26). Estimation and filtering by reversible jump MCMC for a doubly stochastic Poisson model for ultra-high-frequency financial data. Statistical Modelling, 6, S. Centanni and M. Minozzo (26). A Monte Carlo approach to filtering for a class of marked doubly stochastic Poisson processes. Journal of the American Statistical Association,, R. Engle and J. Russell (998). Autoregressive conditional duration: A new model of irregularlt spaced transaction data. Econometrica, 66, R. Frey (2). Risk-minimization with incomplete information in a model for high-frequency data. Mathematical Finance,, R. Frey and W. Runggaldier (2). A nonlinear filtering approach to volatility estimation with a view towards high frequency data. International Journal of Theoretical and Applied Finance, 4, 27-3.

79 Bibliography 5.Conclusions Bibliography A. Hall and N. Hautsch (24). A continuous-time measurement of the buy-sell pressure in a limit order book market. Discussion Paper, 4-7, Department of Economics, University of Copenhagen, Copenhagen. A. Hall and N. Hautsch (26). Order aggressiveness and order book dynamics. Empirical Economics, 3, N. Hautsch (25). The latent factor VAR model: testing for a common component in the intraday trading process. Discussion Paper, Finance Research Unit, Department of Economics, University of Copenhagen, Copenhagen. T. Rydberg and N. Shephard (2). A modelling framework for the prices and times of trades made on the New York stock exchange. Cambridge University Press, M. Zhang, J. Russell and R. Tsay (2). A nonlinear autoregressive conditional duration model with applications to financial transaction data. Journal of Econometrics, 4,

Location Multiplicative Error Model. Asymptotic Inference and Empirical Analysis

Location Multiplicative Error Model. Asymptotic Inference and Empirical Analysis : Asymptotic Inference and Empirical Analysis Qian Li Department of Mathematics and Statistics University of Missouri-Kansas City ql35d@mail.umkc.edu October 29, 2015 Outline of Topics Introduction GARCH

More information

arxiv: v1 [q-fin.st] 18 Sep 2017

arxiv: v1 [q-fin.st] 18 Sep 2017 A new approach to the modeling of financial volumes arxiv:179.583v1 [q-fin.st] 18 Sep 17 Guglielmo D Amico Department of Pharmacy, University G. d Annunzio of Chieti-Pescara e-mail: g.damico@unich.it Filippo

More information

Generalized Autoregressive Score Models

Generalized Autoregressive Score Models Generalized Autoregressive Score Models by: Drew Creal, Siem Jan Koopman, André Lucas To capture the dynamic behavior of univariate and multivariate time series processes, we can allow parameters to be

More information

ECONOMICS 7200 MODERN TIME SERIES ANALYSIS Econometric Theory and Applications

ECONOMICS 7200 MODERN TIME SERIES ANALYSIS Econometric Theory and Applications ECONOMICS 7200 MODERN TIME SERIES ANALYSIS Econometric Theory and Applications Yongmiao Hong Department of Economics & Department of Statistical Sciences Cornell University Spring 2019 Time and uncertainty

More information

Accounting for Missing Values in Score- Driven Time-Varying Parameter Models

Accounting for Missing Values in Score- Driven Time-Varying Parameter Models TI 2016-067/IV Tinbergen Institute Discussion Paper Accounting for Missing Values in Score- Driven Time-Varying Parameter Models André Lucas Anne Opschoor Julia Schaumburg Faculty of Economics and Business

More information

Subsampling Cumulative Covariance Estimator

Subsampling Cumulative Covariance Estimator Subsampling Cumulative Covariance Estimator Taro Kanatani Faculty of Economics, Shiga University 1-1-1 Bamba, Hikone, Shiga 5228522, Japan February 2009 Abstract In this paper subsampling Cumulative Covariance

More information

Generalized Autoregressive Conditional Intensity Models with Long Range Dependence

Generalized Autoregressive Conditional Intensity Models with Long Range Dependence Generalized Autoregressive Conditional Intensity Models with Long Range Dependence Nikolaus Hautsch Klausurtagung SFB 649 Ökonomisches Risiko Motzen 15. Juni 2007 Nikolaus Hautsch (HU Berlin) GLMACI Models

More information

Brief introduction to Markov Chain Monte Carlo

Brief introduction to Markov Chain Monte Carlo Brief introduction to Department of Probability and Mathematical Statistics seminar Stochastic modeling in economics and finance November 7, 2011 Brief introduction to Content 1 and motivation Classical

More information

Econ 423 Lecture Notes: Additional Topics in Time Series 1

Econ 423 Lecture Notes: Additional Topics in Time Series 1 Econ 423 Lecture Notes: Additional Topics in Time Series 1 John C. Chao April 25, 2017 1 These notes are based in large part on Chapter 16 of Stock and Watson (2011). They are for instructional purposes

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

The Realized RSDC model

The Realized RSDC model The Realized RSDC model Denis Pelletier North Carolina State University and Aymard Kassi North Carolina State University Current version: March 25, 24 Incomplete and early draft. Abstract We introduce

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

Markov Chain Monte Carlo (MCMC)

Markov Chain Monte Carlo (MCMC) Markov Chain Monte Carlo (MCMC Dependent Sampling Suppose we wish to sample from a density π, and we can evaluate π as a function but have no means to directly generate a sample. Rejection sampling can

More information

Contagious default: application of methods of Statistical Mechanics in Finance

Contagious default: application of methods of Statistical Mechanics in Finance Contagious default: application of methods of Statistical Mechanics in Finance Wolfgang J. Runggaldier University of Padova, Italy www.math.unipd.it/runggaldier based on joint work with : Paolo Dai Pra,

More information

Calibration of Stochastic Volatility Models using Particle Markov Chain Monte Carlo Methods

Calibration of Stochastic Volatility Models using Particle Markov Chain Monte Carlo Methods Calibration of Stochastic Volatility Models using Particle Markov Chain Monte Carlo Methods Jonas Hallgren 1 1 Department of Mathematics KTH Royal Institute of Technology Stockholm, Sweden BFS 2012 June

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

Frequencies in Ultra-high-frequent Trading

Frequencies in Ultra-high-frequent Trading Frequencies in Ultra-high-frequent Trading Wing Lon Ng University of Muenster Institute of Econometrics and Economic Statistics E-mail: 05wing@wiwi.uni muenster.de January 8, 2007 Abstract This paper suggests

More information

Exercises Tutorial at ICASSP 2016 Learning Nonlinear Dynamical Models Using Particle Filters

Exercises Tutorial at ICASSP 2016 Learning Nonlinear Dynamical Models Using Particle Filters Exercises Tutorial at ICASSP 216 Learning Nonlinear Dynamical Models Using Particle Filters Andreas Svensson, Johan Dahlin and Thomas B. Schön March 18, 216 Good luck! 1 [Bootstrap particle filter for

More information

SUPPLEMENT TO MARKET ENTRY COSTS, PRODUCER HETEROGENEITY, AND EXPORT DYNAMICS (Econometrica, Vol. 75, No. 3, May 2007, )

SUPPLEMENT TO MARKET ENTRY COSTS, PRODUCER HETEROGENEITY, AND EXPORT DYNAMICS (Econometrica, Vol. 75, No. 3, May 2007, ) Econometrica Supplementary Material SUPPLEMENT TO MARKET ENTRY COSTS, PRODUCER HETEROGENEITY, AND EXPORT DYNAMICS (Econometrica, Vol. 75, No. 3, May 2007, 653 710) BY SANGHAMITRA DAS, MARK ROBERTS, AND

More information

Switching Regime Estimation

Switching Regime Estimation Switching Regime Estimation Series de Tiempo BIrkbeck March 2013 Martin Sola (FE) Markov Switching models 01/13 1 / 52 The economy (the time series) often behaves very different in periods such as booms

More information

Non-homogeneous Markov Mixture of Periodic Autoregressions for the Analysis of Air Pollution in the Lagoon of Venice

Non-homogeneous Markov Mixture of Periodic Autoregressions for the Analysis of Air Pollution in the Lagoon of Venice Non-homogeneous Markov Mixture of Periodic Autoregressions for the Analysis of Air Pollution in the Lagoon of Venice Roberta Paroli 1, Silvia Pistollato, Maria Rosa, and Luigi Spezia 3 1 Istituto di Statistica

More information

Financial Econometrics

Financial Econometrics Financial Econometrics Nonlinear time series analysis Gerald P. Dwyer Trinity College, Dublin January 2016 Outline 1 Nonlinearity Does nonlinearity matter? Nonlinear models Tests for nonlinearity Forecasting

More information

Session 5B: A worked example EGARCH model

Session 5B: A worked example EGARCH model Session 5B: A worked example EGARCH model John Geweke Bayesian Econometrics and its Applications August 7, worked example EGARCH model August 7, / 6 EGARCH Exponential generalized autoregressive conditional

More information

Introduction to Machine Learning CMU-10701

Introduction to Machine Learning CMU-10701 Introduction to Machine Learning CMU-10701 Markov Chain Monte Carlo Methods Barnabás Póczos & Aarti Singh Contents Markov Chain Monte Carlo Methods Goal & Motivation Sampling Rejection Importance Markov

More information

6. The econometrics of Financial Markets: Empirical Analysis of Financial Time Series. MA6622, Ernesto Mordecki, CityU, HK, 2006.

6. The econometrics of Financial Markets: Empirical Analysis of Financial Time Series. MA6622, Ernesto Mordecki, CityU, HK, 2006. 6. The econometrics of Financial Markets: Empirical Analysis of Financial Time Series MA6622, Ernesto Mordecki, CityU, HK, 2006. References for Lecture 5: Quantitative Risk Management. A. McNeil, R. Frey,

More information

Multivariate GARCH models.

Multivariate GARCH models. Multivariate GARCH models. Financial market volatility moves together over time across assets and markets. Recognizing this commonality through a multivariate modeling framework leads to obvious gains

More information

Markov Chain Monte Carlo

Markov Chain Monte Carlo Markov Chain Monte Carlo Recall: To compute the expectation E ( h(y ) ) we use the approximation E(h(Y )) 1 n n h(y ) t=1 with Y (1),..., Y (n) h(y). Thus our aim is to sample Y (1),..., Y (n) from f(y).

More information

A simple graphical method to explore tail-dependence in stock-return pairs

A simple graphical method to explore tail-dependence in stock-return pairs A simple graphical method to explore tail-dependence in stock-return pairs Klaus Abberger, University of Konstanz, Germany Abstract: For a bivariate data set the dependence structure can not only be measured

More information

Determining and Forecasting High-Frequency Value-at-Risk by Using Lévy Processes

Determining and Forecasting High-Frequency Value-at-Risk by Using Lévy Processes Determining and Forecasting High-Frequency Value-at-Risk by Using Lévy Processes W ei Sun 1, Svetlozar Rachev 1,2, F rank J. F abozzi 3 1 Institute of Statistics and Mathematical Economics, University

More information

High-dimensional Problems in Finance and Economics. Thomas M. Mertens

High-dimensional Problems in Finance and Economics. Thomas M. Mertens High-dimensional Problems in Finance and Economics Thomas M. Mertens NYU Stern Risk Economics Lab April 17, 2012 1 / 78 Motivation Many problems in finance and economics are high dimensional. Dynamic Optimization:

More information

Bayesian model selection in graphs by using BDgraph package

Bayesian model selection in graphs by using BDgraph package Bayesian model selection in graphs by using BDgraph package A. Mohammadi and E. Wit March 26, 2013 MOTIVATION Flow cytometry data with 11 proteins from Sachs et al. (2005) RESULT FOR CELL SIGNALING DATA

More information

Bayesian Semiparametric GARCH Models

Bayesian Semiparametric GARCH Models Bayesian Semiparametric GARCH Models Xibin (Bill) Zhang and Maxwell L. King Department of Econometrics and Business Statistics Faculty of Business and Economics xibin.zhang@monash.edu Quantitative Methods

More information

Computational statistics

Computational statistics Computational statistics Markov Chain Monte Carlo methods Thierry Denœux March 2017 Thierry Denœux Computational statistics March 2017 1 / 71 Contents of this chapter When a target density f can be evaluated

More information

Bayesian Semiparametric GARCH Models

Bayesian Semiparametric GARCH Models Bayesian Semiparametric GARCH Models Xibin (Bill) Zhang and Maxwell L. King Department of Econometrics and Business Statistics Faculty of Business and Economics xibin.zhang@monash.edu Quantitative Methods

More information

Marginal Specifications and a Gaussian Copula Estimation

Marginal Specifications and a Gaussian Copula Estimation Marginal Specifications and a Gaussian Copula Estimation Kazim Azam Abstract Multivariate analysis involving random variables of different type like count, continuous or mixture of both is frequently required

More information

Online appendix to On the stability of the excess sensitivity of aggregate consumption growth in the US

Online appendix to On the stability of the excess sensitivity of aggregate consumption growth in the US Online appendix to On the stability of the excess sensitivity of aggregate consumption growth in the US Gerdie Everaert 1, Lorenzo Pozzi 2, and Ruben Schoonackers 3 1 Ghent University & SHERPPA 2 Erasmus

More information

Bayesian Inference for Pair-copula Constructions of Multiple Dependence

Bayesian Inference for Pair-copula Constructions of Multiple Dependence Bayesian Inference for Pair-copula Constructions of Multiple Dependence Claudia Czado and Aleksey Min Technische Universität München cczado@ma.tum.de, aleksmin@ma.tum.de December 7, 2007 Overview 1 Introduction

More information

Quantitative Methods in High-Frequency Financial Econometrics:Modeling Univariate and Multivariate Time Series

Quantitative Methods in High-Frequency Financial Econometrics:Modeling Univariate and Multivariate Time Series Quantitative Methods in High-Frequency Financial Econometrics:Modeling Univariate and Multivariate Time Series W ei Sun Institute of Statistics and Mathematical Economics, University of Karlsruhe, Germany

More information

A Guide to Modern Econometric:

A Guide to Modern Econometric: A Guide to Modern Econometric: 4th edition Marno Verbeek Rotterdam School of Management, Erasmus University, Rotterdam B 379887 )WILEY A John Wiley & Sons, Ltd., Publication Contents Preface xiii 1 Introduction

More information

Bayesian Extreme Quantile Regression for Hidden Markov Models

Bayesian Extreme Quantile Regression for Hidden Markov Models Bayesian Extreme Quantile Regression for Hidden Markov Models A thesis submitted for the degree of Doctor of Philosophy by Antonios Koutsourelis Supervised by Dr. Keming Yu and Dr. Antoaneta Serguieva

More information

Reminder of some Markov Chain properties:

Reminder of some Markov Chain properties: Reminder of some Markov Chain properties: 1. a transition from one state to another occurs probabilistically 2. only state that matters is where you currently are (i.e. given present, future is independent

More information

13. Estimation and Extensions in the ARCH model. MA6622, Ernesto Mordecki, CityU, HK, References for this Lecture:

13. Estimation and Extensions in the ARCH model. MA6622, Ernesto Mordecki, CityU, HK, References for this Lecture: 13. Estimation and Extensions in the ARCH model MA6622, Ernesto Mordecki, CityU, HK, 2006. References for this Lecture: Robert F. Engle. GARCH 101: The Use of ARCH/GARCH Models in Applied Econometrics,

More information

Sample of Ph.D. Advisory Exam For MathFinance

Sample of Ph.D. Advisory Exam For MathFinance Sample of Ph.D. Advisory Exam For MathFinance Students who wish to enter the Ph.D. program of Mathematics of Finance are required to take the advisory exam. This exam consists of three major parts. The

More information

SAMSI Astrostatistics Tutorial. More Markov chain Monte Carlo & Demo of Mathematica software

SAMSI Astrostatistics Tutorial. More Markov chain Monte Carlo & Demo of Mathematica software SAMSI Astrostatistics Tutorial More Markov chain Monte Carlo & Demo of Mathematica software Phil Gregory University of British Columbia 26 Bayesian Logical Data Analysis for the Physical Sciences Contents:

More information

A Semiparametric Conditional Duration Model

A Semiparametric Conditional Duration Model A Semiparametric Conditional Duration Model Mardi Dungey y Xiangdong Long z Aman Ullah x Yun Wang { April, 01 ABSTRACT We propose a new semiparametric autoregressive duration (SACD) model, which incorporates

More information

Research Article Optimal Portfolio Estimation for Dependent Financial Returns with Generalized Empirical Likelihood

Research Article Optimal Portfolio Estimation for Dependent Financial Returns with Generalized Empirical Likelihood Advances in Decision Sciences Volume 2012, Article ID 973173, 8 pages doi:10.1155/2012/973173 Research Article Optimal Portfolio Estimation for Dependent Financial Returns with Generalized Empirical Likelihood

More information

ST 740: Markov Chain Monte Carlo

ST 740: Markov Chain Monte Carlo ST 740: Markov Chain Monte Carlo Alyson Wilson Department of Statistics North Carolina State University October 14, 2012 A. Wilson (NCSU Stsatistics) MCMC October 14, 2012 1 / 20 Convergence Diagnostics:

More information

Bootstrap tests of multiple inequality restrictions on variance ratios

Bootstrap tests of multiple inequality restrictions on variance ratios Economics Letters 91 (2006) 343 348 www.elsevier.com/locate/econbase Bootstrap tests of multiple inequality restrictions on variance ratios Jeff Fleming a, Chris Kirby b, *, Barbara Ostdiek a a Jones Graduate

More information

eqr094: Hierarchical MCMC for Bayesian System Reliability

eqr094: Hierarchical MCMC for Bayesian System Reliability eqr094: Hierarchical MCMC for Bayesian System Reliability Alyson G. Wilson Statistical Sciences Group, Los Alamos National Laboratory P.O. Box 1663, MS F600 Los Alamos, NM 87545 USA Phone: 505-667-9167

More information

Duration-Based Volatility Estimation

Duration-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 information

Stat 535 C - Statistical Computing & Monte Carlo Methods. Lecture 18-16th March Arnaud Doucet

Stat 535 C - Statistical Computing & Monte Carlo Methods. Lecture 18-16th March Arnaud Doucet Stat 535 C - Statistical Computing & Monte Carlo Methods Lecture 18-16th March 2006 Arnaud Doucet Email: arnaud@cs.ubc.ca 1 1.1 Outline Trans-dimensional Markov chain Monte Carlo. Bayesian model for autoregressions.

More information

Kernel Sequential Monte Carlo

Kernel Sequential Monte Carlo Kernel Sequential Monte Carlo Ingmar Schuster (Paris Dauphine) Heiko Strathmann (University College London) Brooks Paige (Oxford) Dino Sejdinovic (Oxford) * equal contribution April 25, 2016 1 / 37 Section

More information

A quick introduction to Markov chains and Markov chain Monte Carlo (revised version)

A quick introduction to Markov chains and Markov chain Monte Carlo (revised version) A quick introduction to Markov chains and Markov chain Monte Carlo (revised version) Rasmus Waagepetersen Institute of Mathematical Sciences Aalborg University 1 Introduction These notes are intended to

More information

Paul Karapanagiotidis ECO4060

Paul Karapanagiotidis ECO4060 Paul Karapanagiotidis ECO4060 The way forward 1) Motivate why Markov-Chain Monte Carlo (MCMC) is useful for econometric modeling 2) Introduce Markov-Chain Monte Carlo (MCMC) - Metropolis-Hastings (MH)

More information

Order book modeling and market making under uncertainty.

Order book modeling and market making under uncertainty. Order book modeling and market making under uncertainty. Sidi Mohamed ALY Lund University sidi@maths.lth.se (Joint work with K. Nyström and C. Zhang, Uppsala University) Le Mans, June 29, 2016 1 / 22 Outline

More information

16 : Approximate Inference: Markov Chain Monte Carlo

16 : Approximate Inference: Markov Chain Monte Carlo 10-708: Probabilistic Graphical Models 10-708, Spring 2017 16 : Approximate Inference: Markov Chain Monte Carlo Lecturer: Eric P. Xing Scribes: Yuan Yang, Chao-Ming Yen 1 Introduction As the target distribution

More information

Monte Carlo Methods. Leon Gu CSD, CMU

Monte Carlo Methods. Leon Gu CSD, CMU Monte Carlo Methods Leon Gu CSD, CMU Approximate Inference EM: y-observed variables; x-hidden variables; θ-parameters; E-step: q(x) = p(x y, θ t 1 ) M-step: θ t = arg max E q(x) [log p(y, x θ)] θ Monte

More information

Generalized normal mean variance mixtures and subordinated Brownian motions in Finance. Elisa Luciano and Patrizia Semeraro

Generalized normal mean variance mixtures and subordinated Brownian motions in Finance. Elisa Luciano and Patrizia Semeraro Generalized normal mean variance mixtures and subordinated Brownian motions in Finance. Elisa Luciano and Patrizia Semeraro Università degli Studi di Torino Dipartimento di Statistica e Matematica Applicata

More information

Doubly Inhomogeneous Cluster Point Processes

Doubly Inhomogeneous Cluster Point Processes Doubly Inhomogeneous Cluster Point Processes Tomáš Mrkvička, Samuel Soubeyrand May 2016 Abscissa Motivation - Group dispersal model It is dispersal model, where particles are released in groups by a single

More information

Finite Sample Analysis of Weighted Realized Covariance with Noisy Asynchronous Observations

Finite Sample Analysis of Weighted Realized Covariance with Noisy Asynchronous Observations Finite Sample Analysis of Weighted Realized Covariance with Noisy Asynchronous Observations Taro Kanatani JSPS Fellow, Institute of Economic Research, Kyoto University October 18, 2007, Stanford 1 Introduction

More information

New stylized facts in financial markets: The Omori law and price impact of a single transaction in financial markets

New stylized facts in financial markets: The Omori law and price impact of a single transaction in financial markets Observatory of Complex Systems, Palermo, Italy Rosario N. Mantegna New stylized facts in financial markets: The Omori law and price impact of a single transaction in financial markets work done in collaboration

More information

Computer Vision Group Prof. Daniel Cremers. 11. Sampling Methods: Markov Chain Monte Carlo

Computer Vision Group Prof. Daniel Cremers. 11. Sampling Methods: Markov Chain Monte Carlo Group Prof. Daniel Cremers 11. Sampling Methods: Markov Chain Monte Carlo Markov Chain Monte Carlo In high-dimensional spaces, rejection sampling and importance sampling are very inefficient An alternative

More information

Hastings-within-Gibbs Algorithm: Introduction and Application on Hierarchical Model

Hastings-within-Gibbs Algorithm: Introduction and Application on Hierarchical Model UNIVERSITY OF TEXAS AT SAN ANTONIO Hastings-within-Gibbs Algorithm: Introduction and Application on Hierarchical Model Liang Jing April 2010 1 1 ABSTRACT In this paper, common MCMC algorithms are introduced

More information

Sample Exam Questions for Econometrics

Sample Exam Questions for Econometrics Sample Exam Questions for Econometrics 1 a) What is meant by marginalisation and conditioning in the process of model reduction within the dynamic modelling tradition? (30%) b) Having derived a model for

More information

Riemann Manifold Methods in Bayesian Statistics

Riemann Manifold Methods in Bayesian Statistics Ricardo Ehlers ehlers@icmc.usp.br Applied Maths and Stats University of São Paulo, Brazil Working Group in Statistical Learning University College Dublin September 2015 Bayesian inference is based on Bayes

More information

18 : Advanced topics in MCMC. 1 Gibbs Sampling (Continued from the last lecture)

18 : Advanced topics in MCMC. 1 Gibbs Sampling (Continued from the last lecture) 10-708: Probabilistic Graphical Models 10-708, Spring 2014 18 : Advanced topics in MCMC Lecturer: Eric P. Xing Scribes: Jessica Chemali, Seungwhan Moon 1 Gibbs Sampling (Continued from the last lecture)

More information

Long Memory through Marginalization

Long Memory through Marginalization Long Memory through Marginalization Hidden & Ignored Cross-Section Dependence Guillaume Chevillon ESSEC Business School (CREAR) and CREST joint with Alain Hecq and Sébastien Laurent Maastricht University

More information

Hmms with variable dimension structures and extensions

Hmms with variable dimension structures and extensions Hmm days/enst/january 21, 2002 1 Hmms with variable dimension structures and extensions Christian P. Robert Université Paris Dauphine www.ceremade.dauphine.fr/ xian Hmm days/enst/january 21, 2002 2 1 Estimating

More information

DEPARTMENT OF ECONOMETRICS AND BUSINESS STATISTICS

DEPARTMENT OF ECONOMETRICS AND BUSINESS STATISTICS ISSN 1440-771X AUSTRALIA DEPARTMENT OF ECONOMETRICS AND BUSINESS STATISTICS Bayesian Analysis of the Stochastic Conditional Duration Model Chris M. Strickland, Catherine S. Forbes and Gael M. Martin Working

More information

Intro VEC and BEKK Example Factor Models Cond Var and Cor Application Ref 4. MGARCH

Intro VEC and BEKK Example Factor Models Cond Var and Cor Application Ref 4. MGARCH ntro VEC and BEKK Example Factor Models Cond Var and Cor Application Ref 4. MGARCH JEM 140: Quantitative Multivariate Finance ES, Charles University, Prague Summer 2018 JEM 140 () 4. MGARCH Summer 2018

More information

Time Series Models for Measuring Market Risk

Time Series Models for Measuring Market Risk Time Series Models for Measuring Market Risk José Miguel Hernández Lobato Universidad Autónoma de Madrid, Computer Science Department June 28, 2007 1/ 32 Outline 1 Introduction 2 Competitive and collaborative

More information

CONTAGION VERSUS FLIGHT TO QUALITY IN FINANCIAL MARKETS

CONTAGION VERSUS FLIGHT TO QUALITY IN FINANCIAL MARKETS EVA IV, CONTAGION VERSUS FLIGHT TO QUALITY IN FINANCIAL MARKETS Jose Olmo Department of Economics City University, London (joint work with Jesús Gonzalo, Universidad Carlos III de Madrid) 4th Conference

More information

Markov Chain Monte Carlo

Markov Chain Monte Carlo Chapter 5 Markov Chain Monte Carlo MCMC is a kind of improvement of the Monte Carlo method By sampling from a Markov chain whose stationary distribution is the desired sampling distributuion, it is possible

More information

Bayesian Model Comparison:

Bayesian Model Comparison: Bayesian Model Comparison: Modeling Petrobrás log-returns Hedibert Freitas Lopes February 2014 Log price: y t = log p t Time span: 12/29/2000-12/31/2013 (n = 3268 days) LOG PRICE 1 2 3 4 0 500 1000 1500

More information

Research Division Federal Reserve Bank of St. Louis Working Paper Series

Research Division Federal Reserve Bank of St. Louis Working Paper Series Research Division Federal Reserve Bank of St Louis Working Paper Series Kalman Filtering with Truncated Normal State Variables for Bayesian Estimation of Macroeconomic Models Michael Dueker Working Paper

More information

GARCH Models. Eduardo Rossi University of Pavia. December Rossi GARCH Financial Econometrics / 50

GARCH Models. Eduardo Rossi University of Pavia. December Rossi GARCH Financial Econometrics / 50 GARCH Models Eduardo Rossi University of Pavia December 013 Rossi GARCH Financial Econometrics - 013 1 / 50 Outline 1 Stylized Facts ARCH model: definition 3 GARCH model 4 EGARCH 5 Asymmetric Models 6

More information

Convex Optimization CMU-10725

Convex Optimization CMU-10725 Convex Optimization CMU-10725 Simulated Annealing Barnabás Póczos & Ryan Tibshirani Andrey Markov Markov Chains 2 Markov Chains Markov chain: Homogen Markov chain: 3 Markov Chains Assume that the state

More information

Bayesian Methods for Machine Learning

Bayesian Methods for Machine Learning Bayesian Methods for Machine Learning CS 584: Big Data Analytics Material adapted from Radford Neal s tutorial (http://ftp.cs.utoronto.ca/pub/radford/bayes-tut.pdf), Zoubin Ghahramni (http://hunch.net/~coms-4771/zoubin_ghahramani_bayesian_learning.pdf),

More information

Empirical properties of large covariance matrices in finance

Empirical properties of large covariance matrices in finance Empirical properties of large covariance matrices in finance Ex: RiskMetrics Group, Geneva Since 2010: Swissquote, Gland December 2009 Covariance and large random matrices Many problems in finance require

More information

Markov Chain Monte Carlo methods

Markov Chain Monte Carlo methods Markov Chain Monte Carlo methods By Oleg Makhnin 1 Introduction a b c M = d e f g h i 0 f(x)dx 1.1 Motivation 1.1.1 Just here Supresses numbering 1.1.2 After this 1.2 Literature 2 Method 2.1 New math As

More information

Bayesian Estimation of Input Output Tables for Russia

Bayesian Estimation of Input Output Tables for Russia Bayesian Estimation of Input Output Tables for Russia Oleg Lugovoy (EDF, RANE) Andrey Polbin (RANE) Vladimir Potashnikov (RANE) WIOD Conference April 24, 2012 Groningen Outline Motivation Objectives Bayesian

More information

Markov chain Monte Carlo

Markov chain Monte Carlo Markov chain Monte Carlo Feng Li feng.li@cufe.edu.cn School of Statistics and Mathematics Central University of Finance and Economics Revised on April 24, 2017 Today we are going to learn... 1 Markov Chains

More information

Portfolio Allocation using High Frequency Data. Jianqing Fan

Portfolio Allocation using High Frequency Data. Jianqing Fan Portfolio Allocation using High Frequency Data Princeton University With Yingying Li and Ke Yu http://www.princeton.edu/ jqfan September 10, 2010 About this talk How to select sparsely optimal portfolio?

More information

Continuous Time Approximations to GARCH(1, 1)-Family Models and Their Limiting Properties

Continuous Time Approximations to GARCH(1, 1)-Family Models and Their Limiting Properties Communications for Statistical Applications and Methods 214, Vol. 21, No. 4, 327 334 DOI: http://dx.doi.org/1.5351/csam.214.21.4.327 Print ISSN 2287-7843 / Online ISSN 2383-4757 Continuous Time Approximations

More information

Do Markov-Switching Models Capture Nonlinearities in the Data? Tests using Nonparametric Methods

Do Markov-Switching Models Capture Nonlinearities in the Data? Tests using Nonparametric Methods Do Markov-Switching Models Capture Nonlinearities in the Data? Tests using Nonparametric Methods Robert V. Breunig Centre for Economic Policy Research, Research School of Social Sciences and School of

More information

Variance stabilization and simple GARCH models. Erik Lindström

Variance stabilization and simple GARCH models. Erik Lindström Variance stabilization and simple GARCH models Erik Lindström Simulation, GBM Standard model in math. finance, the GBM ds t = µs t dt + σs t dw t (1) Solution: S t = S 0 exp ) ) ((µ σ2 t + σw t 2 (2) Problem:

More information

Pseudo-marginal MCMC methods for inference in latent variable models

Pseudo-marginal MCMC methods for inference in latent variable models Pseudo-marginal MCMC methods for inference in latent variable models Arnaud Doucet Department of Statistics, Oxford University Joint work with George Deligiannidis (Oxford) & Mike Pitt (Kings) MCQMC, 19/08/2016

More information

Computer Intensive Methods in Mathematical Statistics

Computer Intensive Methods in Mathematical Statistics Computer Intensive Methods in Mathematical Statistics Department of mathematics johawes@kth.se Lecture 16 Advanced topics in computational statistics 18 May 2017 Computer Intensive Methods (1) Plan of

More information

Markov Chain Monte Carlo Lecture 6

Markov Chain Monte Carlo Lecture 6 Sequential parallel tempering With the development of science and technology, we more and more need to deal with high dimensional systems. For example, we need to align a group of protein or DNA sequences

More information

Lecture 4: Dynamic models

Lecture 4: Dynamic models linear s Lecture 4: s Hedibert Freitas Lopes The University of Chicago Booth School of Business 5807 South Woodlawn Avenue, Chicago, IL 60637 http://faculty.chicagobooth.edu/hedibert.lopes hlopes@chicagobooth.edu

More information

Control Variates for Markov Chain Monte Carlo

Control Variates for Markov Chain Monte Carlo Control Variates for Markov Chain Monte Carlo Dellaportas, P., Kontoyiannis, I., and Tsourti, Z. Dept of Statistics, AUEB Dept of Informatics, AUEB 1st Greek Stochastics Meeting Monte Carlo: Probability

More information

Graduate Econometrics I: What is econometrics?

Graduate Econometrics I: What is econometrics? Graduate Econometrics I: What is econometrics? Yves Dominicy Université libre de Bruxelles Solvay Brussels School of Economics and Management ECARES Yves Dominicy Graduate Econometrics I: What is econometrics?

More information

Multivariate Stress Scenarios and Solvency

Multivariate Stress Scenarios and Solvency Multivariate Stress Scenarios and Solvency Alexander J. McNeil 1 1 Heriot-Watt University Edinburgh Croatian Quants Day Zagreb 11th May 2012 a.j.mcneil@hw.ac.uk AJM Stress Testing 1 / 51 Regulation General

More information

Learning Static Parameters in Stochastic Processes

Learning Static Parameters in Stochastic Processes Learning Static Parameters in Stochastic Processes Bharath Ramsundar December 14, 2012 1 Introduction Consider a Markovian stochastic process X T evolving (perhaps nonlinearly) over time variable T. We

More information

Modelling Trades-Through in a Limit Order Book Using Hawkes Processes

Modelling Trades-Through in a Limit Order Book Using Hawkes Processes Vol. 6, 212-22 June 14, 212 http://dx.doi.org/1.518/economics-ejournal.ja.212-22 Modelling Trades-Through in a Limit Order Book Using Hawkes Processes Ioane Muni Toke Ecole Centrale Paris and University

More information

Diagnostic Test for GARCH Models Based on Absolute Residual Autocorrelations

Diagnostic Test for GARCH Models Based on Absolute Residual Autocorrelations Diagnostic Test for GARCH Models Based on Absolute Residual Autocorrelations Farhat Iqbal Department of Statistics, University of Balochistan Quetta-Pakistan farhatiqb@gmail.com Abstract In this paper

More information

FORECASTING IN FINANCIAL MARKET USING MARKOV R MARKOV REGIME SWITCHING AND PRINCIPAL COMPONENT ANALYSIS.

FORECASTING IN FINANCIAL MARKET USING MARKOV R MARKOV REGIME SWITCHING AND PRINCIPAL COMPONENT ANALYSIS. FORECASTING IN FINANCIAL MARKET USING MARKOV REGIME SWITCHING AND PRINCIPAL COMPONENT ANALYSIS. September 13, 2012 1 2 3 4 5 Heteroskedasticity Model Multicollinearily 6 In Thai Outline การพยากรณ ในตลาดทางการเง

More information

Testing Monotonicity of Pricing Kernels

Testing Monotonicity of Pricing Kernels Yuri Golubev Wolfgang Härdle Roman Timofeev C.A.S.E. Center for Applied Statistics and Economics Humboldt-Universität zu Berlin 12 1 8 6 4 2 2 4 25 2 15 1 5 5 1 15 2 25 Motivation 2-2 Motivation An investor

More information

Multilevel Statistical Models: 3 rd edition, 2003 Contents

Multilevel Statistical Models: 3 rd edition, 2003 Contents Multilevel Statistical Models: 3 rd edition, 2003 Contents Preface Acknowledgements Notation Two and three level models. A general classification notation and diagram Glossary Chapter 1 An introduction

More information

Markov Chains and MCMC

Markov Chains and MCMC Markov Chains and MCMC Markov chains Let S = {1, 2,..., N} be a finite set consisting of N states. A Markov chain Y 0, Y 1, Y 2,... is a sequence of random variables, with Y t S for all points in time

More information