Auxiliary Results for Activity Signature Functions
|
|
- Sybil Thomas
- 5 years ago
- Views:
Transcription
1 Auxiliary esults for Activity Sigature Fuctios Viktor Todorov ad George Tauche Jue 9, 29 Abstract This documet cotais additioal results pertaiig to the paper Activity Sigature Fuctios for High-Frequecy Data Aalysis, Todorov ad Tauche 29). The otatio is the same as i that paper. Departmet of Fiace, Kellogg School of Maagemet, Northwester Uiversity, Evasto, IL 628; v-todorov@kellogg.orthwester.edu Departmet of Ecoomics, Duke Uiversity, Durham, NC 2778; george.tauche@duke.edu. 1
2 Cotets 1 The Effects of Measuremet Error 3 2 Scaled log-power Variatio 3 3 Mote Carlo Mote Carlo Scearios Precisio of the Poit Estimator of Activity Size ad Power of the Test for Cotiuous Martigale Proofs of Theoretical esults Prelimiary esults Limit theorems for realized power variatio i pure-jump case Proof of Theorem 1 i the paper Proof of Theorem 2 i the paper
3 1 The Effects of Measuremet Error We aalyze the effect of measuremet error o our activity sigature fuctio. We will suppose that istead of the true process we observe the process with some error. For example, it is well kow that at very high-frequecy, fiacial data ca be cotamiated by microstructure oise; the literature o microstructure oise is vast, see Bardorff-Nielse ad Shephard 27) for a timely ad extesive review. We will focus oly o the i.i.d. case. The followig theorem gives the behavior of the realized power variatio i this situatio. Theorem 1 Let ɛ i ) i be a i.i.d. sequece of radom variables with fiite momets ad set Υ t = Υ t + ɛ t where Υ is oe of the processes X, Y or Z defied i the paper. The, if assumptio A1 holds, we have V p, Υ, ) t where K p is some costat depedig o p. u.c.p. K p t, As we ca see from the theorem, the asymptotic behavior of the realized power variatio is completely determied by the oise i the price process. Therefore, it is easy to see that for ay p ad Υ = Y, X or Z, β,t ] Υ, p) will diverge to ifiity as we start samplig more frequetly. Ituitively, the measuremet error becomes the most active part of the observed process, sice its magitude does ot decrease as we start samplig more frequetly. Proof of Theorem 1: First, usig a localizatio argumet we ca ad will) assume the stroger assumptio A1. Secod, by a stadard LLN we have [t/ ] i=1 ɛ i ɛ i 1) p P te ɛ u ɛ s p, for u s. Thus, to prove Theorem 1 it suffices to show [t/ V p, Υ, ] ) t ɛ i ɛ i 1) p This follows from i=1 u.c.p.. 1. the elemetary iequality x + y p x p K p x p 1 + y p 1) y, for arbitrary x ad y ad K p beig some costat that depeds o p. 2. the followig well-kow result see Jacod 28)) E i 1 i Z p K p/ applicatio of Hölder iequality. 2 Scaled log-power Variatio A alterative to usig differeces across time scales to determie the activity level is, from Woerer 26), to compute a direct scaled log-power variatio measure β,t ] X, p) = p l ) / l V p, X, k ) T ). 3
4 Scaled Log Power Variatio, Tempered Stable, β= β Figure 1: Scaled log power variatio for the symmetric tempered stable process with β = 1. ad other parameters give by Case TS1. of Table 1; the figures show B Υ, p, α) as a fuctio of p for the quatiles α =.25, α =.5, ad α =.75, readig bottom up. This measure asymptotically behaves as β,t ] X, p) i Sectio 4 of the text; however, Theorem 1 suggests a bias which is apparet i Figure 1. The figure shows three quatile plots of the fuctio B Υ, p, α) := β [αn] p), which is defied aalogously to the quatile fuctio i Sectio 4 of the paper. For the figure, Υ is simulated from case TS1.. Evidetally, B Υ, p, α) is off ceter ad the iterquatile rage is quite large. The bias evidet i Figure 1 is caceled out usig the the two-scale approach. 4
5 3 Mote Carlo 3.1 Mote Carlo Scearios As oted i the paper, some readers, particularly i ecoometrics, prefer poit estimates ad formal test statistics to graphically-based iterval estimates. Therefore, this appedix examies i detail via Mote Carlo the properties of the poit estimator of the activity idex preseted i Sectio 4.3 of the paper. The estimator is a equi-weighted average of the activity sigature fuctio over a data-depedet iterval. Clearly there are may other weightig schemes that could be used to develop estimators of activity, ad explorig all of their properties is far beyod the scope of the paper ad this appedix. ather, our aim is to check if this particular easilycomputed estimator performs reasoably well over a wide variety of estimatio cotexts. We also examie the properties of the formal test of a jump diffusio agaist the alterative of a pure jump model developed i Subsectio 3.3 of the paper. The various scearios goverig the dyamics of the process Z are show i Table 1. The table shows the parameter settigs ad the implied populatio value of the activity idex, deoted β, which is always determied by the domiat compoet of the Z process. Cases A I ad N are Lévy processes. I A C the Lévy process is pure jump while i D it is a cotiuous/browia martigale. I cases E G rare jumps, i.e. compoud Poisso jumps, are added to a cotiuous martigale. Case H represets a particular challege where the activity idex is zero, as does I where the estimator faces the superpositio of two Lévy processes with idices 1. ad 1.5, respectively, with domiat compoet 1.5. Cases J M are Lévy-drive OU or CAMA1, ), with strog persistece ad very little mea reversio, where the backgroud drivig Lévy processes are Cases A D, respectively. Case N is a cotiuous/browia martigale plus a ifiitely active pure jump process. The values of the tuig parameters are set to be realistic for high-frequecy fiacial data ad were determied by kowledge gleamed from earlier studies, so the fidigs below eed to be iterpreted ad qualified accordigly. A very ituitive way to set up the samplig scheme is to thik of the basic time uit, or time period, as oe 24-hour day with observatios take at samplig itervals of 1-mi, 5-mi, 1-mi, ad 3-mi itervals. The first two are typical of may fiacial applicatios while the latter two are relatively coarse. These values correspod to M = 144, 25, 144, 72 observatios per time period, respectively, or samplig itervals of width =.7,.35,.69,.289, which rages over three orders of magitude. Each Mote Carlo pseudo data set is comprised of multiple days worth of data, which we take as N = 25, 15, 3, or 1-year, 5-years, ad 1-years worth of data. Thus M determies the samplig frequecy ad N the spa of the data set, ad the values here are typical of high frequecy fiacial data sets. Of course for a give sceario i Table 1 the outcome oly depeds upo the width of the samplig iterval ad the otios of days ad years are simple labels to guide ituitio. 5
6 Table 1: Mote Carlo Scearios: Dyamics for Z = Y + X Case β ρ σ 2 1 σ 2 2 Jump Specificatio X) A tempered stable with A = 1, β =.5 ad λ =.5 B tempered stable with A = 1, β = 1. ad λ =.5 C tempered stable with A = 1, β = 1.5 ad λ =.5 D oe E rare-jump with λ J =.4, τ = F rare-jump with λ J =.3333, τ =.7746 G rare-jump with λ J = 2., τ =.3162 H variace gamma: tempered stable with A = 1, β =. ad λ =.5 I superpositio of two idepedet tempered stables: oe with A = 1, β = 1.5 ad λ =.5 ad the other with A = 1, β = 1. ad λ =.5 J tempered stable with A = 1, β =.5 ad λ =.5 K tempered stable with A = 1, β = 1. ad λ =.5 L tempered stable with A = 1, β = 1.5 ad λ =.5 M oe N tempered stable with A = 1, β = 1.5 ad λ =.5 Note: With Z = Y + X the cotiuous compoet Y is otrivial if σ1 2 compoet X is otrivial if σ2 2 >. >, ad the jump 6
7 3.2 Precisio of the Poit Estimator of Activity For each sceerio ad M, N) pair we geerated 1 Mote Carlo replicates, computed the poit estimate of β o a period-by-period basis, i.e., day-by-day, ad took the media of the N values as the overall estimate for that replicate. This etails some loss i efficiecy because it does ot impose equality of β across time time periods, but it mimics actual practice where thigs are doe day-by-day to provide some robustess. Across the 1 Mote Carlo replicatios we compute the usual robust summary statistics give by the media, iter-quartile rage, ad media-absolute-deviatio about the the populatio value β. I order to assess the practical accuracy of the estimator, we also computed error rates as the proportio of the 1 replicates that the absolute deviatio of the overall estimator from the populatio value β is o more tha the pre-specified value α =.15,.1,.5. For the activity parameter β [, 2], estimatig it to withi.1 of the true value seems sufficietly accurate for may applicatios, while to withi.15 is a little loose ad.5 very precise. Tables 2 8 displays the summary measures from this Mote Carlo experimet. Oe immediately sees from the iter-quartile rages ad media absolute deviatios that the estimator is very tightly cocetrated aroud a value that is geerally, but ot always close to the true populatio variable. I other words, the samplig variace is early always very small ad the bias geerally small, or modest, but the bias ca be quite large i certai extreme circumstaces. The samplig distributio cocetrates aroud a value that is usually close to the correct value β, but i some circumstaces it cocetrates aroud a value somewhat distat from the correct value. The coclusios from Tables 2 8 are readily summarized. The key cotrol parameter is the samplig frequecy, M, or equivaletly the samplig iterval = 1/M. For samplig correspodig to 1-mi ad 5-mi, the poit estimator is geerally quite accurate across essetially all scearios ad data spas N. O the other had, for samplig at 1-mi ad coarser the estimator ca sometimes be quite off the mark as i Cases E, F, G, ad H, for example. The fact that reducig sharply decreases the bias while icreasig the spa N oly moderately reduces variace is ot surprisig because the asymptotic theory is always of the fill-i type where. A reasoable level of accuracy requires a large, dese, data set of the type that is commoly used i fiacial ecoometrics ad presumably other applicatios as well. Aother poit of ote is that for values of the idex β at or above.5 the estimator is geerally quite accurate at the higher samplig frequecies, while smaller values could require other approaches. Note that i case H β = the estimator cocetrates at the value τ =.1, which is the lower boud of the of iterval used to average the activity sigature fuctio. Keep i mid, however, that if β =, or is close to zero, the the plot of the activity sigature fuctio will reveal just that. 7
8 Table 2: Summary Statistics for 1 eplicates of β Defied i Subsectio 4.3 Case N M β MED IQ MAD Error rates: α-away from the truth α =.15 α =.1 α =.5 A A A A A A A A A A A A B B B B B B B B B B B B Note: N is the umber of uits of time i a simulatio; M is the umber of high-frequecy observatios per uit of time. β is the populatio value. MED is the media, IQ the iter-quartile rage, MAD the media absolute deviatio, each computed over the 1 replicatios. Error rates are the proportio of replicates where ˆβ β > α. 8
9 Table 3: Summary Statistics for 1 eplicates of β Defied i Subsectio 4.3 Case N M β MED IQ MAD Error rates: α-away from the truth α =.15 α =.1 α =.5 C C C C C C C C C C C C D D D D D D D D D D D D Note: N is the umber of uits of time i a simulatio; M is the umber of high-frequecy observatios per uit of time. β is the populatio value. MED is the media, IQ the iter-quartile rage, MAD the media absolute deviatio, each computed over the 1 replicatios. Error rates are the proportio of replicates where β β > α. 9
10 Table 4: Summary Statistics for 1 eplicates of β Defied i Subsectio 4.3 Case N M β MED IQ MAD Error rates: α-away from the truth α =.15 α =.1 α =.5 E E E E E E E E E E E E F F F F F F F F F F F F Note: N is the umber of uits of time i a simulatio; M is the umber of high-frequecy observatios per uit of time. β is the populatio value. MED is the media, IQ the iter-quartile rage, MAD the media absolute deviatio, each computed over the 1 replicatios. Error rates are the proportio of replicates where ˆβ β > α. 1
11 Table 5: Summary Statistics for 1 eplicates of β Defied i Subsectio 4.3 Case N M β MED IQ MAD Error rates: α-away from the truth α =.15 α =.1 α =.5 G G G G G G G G G G G G H H H H H H H H H H H H Note: N is the umber of uits of time i a simulatio; M is the umber of high-frequecy observatios per uit of time. β is the populatio value. MED is the media, IQ the iter-quartile rage, MAD the media absolute deviatio, each computed over the 1 replicatios. Error rates are the proportio of replicates where ˆβ β > α. 11
12 Table 6: Summary Statistics for 1 eplicates of β Defied i Subsectio 4.3 Case N M β MED IQ MAD Error rates: α-away from the truth α =.15 α =.1 α =.5 I I I I I I I I I I I I J J J J J J J J J J J J Note: N is the umber of uits of time i a simulatio; M is the umber of high-frequecy observatios per uit of time. β is the populatio value. MED is the media, IQ the iter-quartile rage, MAD the media absolute deviatio, each computed over the 1 replicatios. Error rates are the proportio of replicates where ˆβ β > α. 12
13 Table 7: Summary Statistics for 1 eplicates of β Defied i Subsectio 4.3 Case N M β MED IQ MAD Error rates: α-away from the truth α =.15 α =.1 α =.5 K K K K K K K K K K K K L L L L L L L L L L L L Note: N is the umber of uits of time i a simulatio; M is the umber of high-frequecy observatios per uit of time. β is the populatio value. MED is the media, IQ the iter-quartile rage, MAD the media absolute deviatio, each computed over the 1 replicatios. Error rates are the proportio of replicates where ˆβ β > α. 13
14 Table 8: Summary Statistics for 1 eplicates of β Defied i Subsectio 4.3 Case N M β MED IQ MAD Error rates: α-away from the truth α =.15 α =.1 α =.5 M M M M M M M M M M M M N N N N N N N N N N N N Note: N is the umber of uits of time i a simulatio; M is the umber of high-frequecy observatios per uit of time. β is the populatio value. MED is the media, IQ the iter-quartile rage, MAD the media absolute deviatio, each computed over the 1 replicatios. Error rates are the proportio of replicates where ˆβ β > α. 14
15 3.3 Size ad Power of the Test for Cotiuous Martigale Tables 9 11 summarize the size ad power properties of the test for a cotiuous martigale proposed i Sectio 3.3 of the paper. The scearios are as i Table 1 above ad the samplig rate M as per Tables 2 8. For each sceario M pair, the test is coducted o each of 1 replicated periods i.e., days mimickig how such tests are coducted i actual practice. The coclusios from these three tables accord well with those of the precedig Sectio 3.2. For the highest samplig frequecy 1-mi) the size ad power properties are excellet, ad with 5-mi samplig the test is oly slightly udersized but with still quite good power properties. O the other had, at the courser samplig itervals there ca be large size distortios ad the test thereby ureliable. The size ad power appear appropriate, at least for samplig frequecies commoly used i fiace. These fidigs very likely carry over to other disciplies where large dese data sets are also available. 15
16 Table 9: Size ad Power of Test for Presece of Cotiuous Martigale Samplig Frequecy Percetage of ejectio of Null Hypothesis α =.1 α =.5 α =.1 Case A H false) M = mi) M = mi) M = mi) M = 48 3 mi) Case B H false) M = mi) M = mi) M = mi) M = 48 3 mi) Case C H false) M = mi) M = mi) M = mi) M = 48 3 mi) Case D H true) M = mi) M = mi) M = mi) M = 48 3 mi) Note: The test is oe-sided ad is based o asymptotic distributio uder the ull of log β t Υ, p)) for p =.9, give i Theorem 3 i the paper. The umber of Mote Carlo replicatios is 1, uits of time ad α deotes the asymptotic) size of the test. 16
17 Table 1: Size ad Power of Test for Presece of Cotiuous Martigale Samplig Frequecy Percetage of ejectio of Null Hypothesis α =.1 α =.5 α =.1 Case E H true) M = mi) M = mi) M = mi) M = 48 3 mi) Case F H true) M = mi) M = mi) M = mi) M = 48 3 mi) Case G H true) M = mi) M = mi) M = mi) M = 48 3 mi) Case H H false) M = mi) M = mi) M = mi) M = 48 3 mi) Case I H false) M = mi) M = mi) M = mi) M = 48 3 mi) Note: The test is oe-sided ad is based o asymptotic distributio uder the ull of log β t Υ, p)) for p =.9, give i Theorem 3 i the paper. The umber of Mote Carlo replicatios is 1, uits of time ad α deotes the asymptotic) size of the test. 17
18 Table 11: Size ad Power of Test for Presece of Cotiuous Martigale Samplig Frequecy Percetage of ejectio of Null Hypothesis α =.1 α =.5 α =.1 Case J H false) M = mi) M = mi) M = mi) M = 48 3 mi) Case K H false) M = mi) M = mi) M = mi) M = 48 3 mi) Case L H false) M = mi) M = mi) M = mi) M = 48 3 mi) Case M H true) M = mi) M = mi) M = mi) M = 48 3 mi) Case N H true) M = mi) M = mi) M = mi) M = 48 3 mi) Note: The test is oe-sided ad is based o asymptotic distributio uder the ull of log β t Υ, p)) for p =.9, give i Theorem 3 i the paper. The umber of Mote Carlo replicatios is 1, uits of time ad α deotes the asymptotic) size of the test. 18
19 4 Proofs of Theoretical esults 4.1 Prelimiary esults First, it is easy to see that the law of the pure-jump model X t, give with equatio 2.2), ca be geerated with the followig process defied o some differet probability space we will still call it X t ) X t = t t + b 2s ds + t + σ 2s κx)1 {y<as } µds, dx, dy) + σ 2s κ x)1 {y<as }µds, dx, dy), 4.1) where ow µ is a Poisso measure but o + + ad with compesator ds νx)dx dy; µds, dx, dy) is the compesated versio of µ. The process X t i 4.1) is writte as a stochastic) itegral with respect to a time-homogeous Poisso radom measure, but we ote that µ is defied o a three dimesioal space. Ituitively, the first dimesio is the time, the secod oe is the size of the jumps but ote that the jumps are multiplied by σ 2t ). The role of the third dimesio is to geerate thiig of the jumps accordig to the process a t ad thus create time-varyig itesity of the jumps i the process X t. I all of the proofs i this sectio we will assume the represetatio i 4.1) for X t, ad we will work with the correspodig probability space ad filtratio which support it. This is coveiet as whe chages we do ot eed to make a chage of the probability space i our proofs. We start with showig some prelimiary results. Deote with L covergece i law o the space D) of oe-dimesioal càdlàg fuctios: +, which is equipped with the Skorokhod topology. We have the followig two Lemmas. Lemma 1 Let Ψ t deote a Lévy process with characteristic triplet b,, F ) with respect to some trucatio fuctio κx), where F is a Lévy measure with F dx) = νx)1 { x ɛ} dx ad νx) is the desity defied i 2.4), while ɛ is a arbitrary positive umber. Further, let b = κx)1 { x ɛ}νx)dx if β < 1 ad b = if β 1. Fially, assume that ν is symmetric whe β = 1. Deote with L t aother Lévy process with characteristic triplet b 1,, F 1 ) with respect to the same trucatio fuctio). b 1 = κ x)ν 1 x)dx if β 1 ad b 1 = κx)ν 1x)dx for β < 1. F 1 is a Lévy measure with F 1 dx) = ν 1 x)dx where ν 1 x) is defied i 2.5). Fially, assume κx) is F 1 -a.s. cotiuous ad i additio that it is symmetric whe β = 1. The as we have: a) b) 1/β Ψ t L L t, 4.2) p/β E Ψ t p ) E L t p ), locally uiformly i t for some p < β. 4.3) Proof: Part a). Sice 1/β Ψ t is a Lévy process to prove the covergece of the sequece we eed to show the covergece of its characteristics, see e.g. Jacod ad Shiryaev 23), Corollary III.3.6. The case β < 1 is straightforward, we show the case β 1. We eed to establish the followig for κ 1/β ) x) 1/β κx) νx)1 { x ɛ} dx κ x)ν 1 x)dx, 4.4) κ 2 1/β x)νx)1 { x ɛ} dx κ 2 x)ν 1 x)dx, 4.5) 19
20 g 1/β x)νx)1 { x ɛ} dx gx)ν 1 x)dx, 4.6) where g is a arbitrary cotiuous ad bouded fuctio o, which is aroud. First, the covergece i 4.5) ad 4.6) trivially follow from a chage of variable i the itegratio ad the fact that the fuctio φ i 2.5) is bouded aroud the origi. Therefore, we are left with showig 4.4). For β = 1 this follows from the symmetry of the trucatio fuctio ad ν. We show 4.4) therefore oly i the case β > 1. First, we have = κ 1/β κ 1/β ) x) 1/β κx) ν 1 x)1 { x ɛ} dx ) x) 1/β x ν 1 x)1 { x ɛ} dx + 1 1/β x κx)) ν 1 x)1 { x ɛ} dx κ x)ν 1 x)dx. 4.7) Thus we will prove 4.4) if we ca show κ 1/β ) x) 1/β κx) ν 2 x)dx. 4.8) It suffices to show 4.4) for κx) = x1{ x 1}. For small eough we have ) 1/β κx) κ 1/β x) ν 2 x)dx = 1 1/β 1 1/β 1 1/β 1 1/β xν 2 x)dx + 1 1/β x β φx)dx + 1 1/β 1/β 1 1/β 1 xν 2 x)dx x β φx)dx. 4.9) For sufficietly small, usig the fact that φx) slowly varies aroud zero, has a fiite limit at ad is itegrable aroud, we have 1 1/β x β φx)dx K + Kφ 1/β ) 1/β β /β, 4.1) for some costat K. The same result obviously holds for 1/β x β φx)dx, ad from here 1 recall we are lookig at the case β > 1) we have the result i 4.8). Thus, combiig 4.7) ad 4.8), we have the result i 4.4) ad therefore 4.2) is established. Part b). First we establish the poitwise covergece for arbitrary t >. Give the result i Part a), we eed oly show that for some p such that p < p < β we have sup E p /β Ψ t p ) <. 4.11) I order to cosider the cases β 1 ad β < 1 together we make a slight abuse of otatio. I what follows we assume that κx) ad tur off the drift b whe β < 1, while whe β 1 we make o chage ad κx) cotiues to be a true trucatio fuctio i.e. it has bouded support ad coicides with the idetity aroud the origi). With this assumptio o κx), deotig with µ the jump measure associated with the process Ψ t ad with µ its compesated versio, we ca write 1/β Ψ t i all cases of β) as 1/β Ψ t = 1/β t κx) µds, dx) + 1/β 2 t κ x)µds, dx). 4.12)
21 The, we ca proceed with the followig decompositio 1/β Ψ t p K 1/β +K +K 1/β 1/β t t t p κx) µds, dx) x >s 1/β p κx) µds, dx) x s 1/β p κ x)µds, dx), 4.13) where K is some costat. Obviously, give the fact that κx) if β < 1, for the first two terms o the right had side of 4.13) we look oly at the case β 1. For the secod itegral o the right-had side of 4.11) for some 1 q 2 such that q > β we have E 1/β K t q/β E p κx) µds, dx) E x s 1/β t κx) q dsνx)dx x s 1/β 1/β )) p /q t κx) µds, dx) x s 1/β q) p /q K, 4.14) where the first iequality follows from the fact that p < q; the secod oe is a applicatio of Burkholder-Davis-Gudy iequality see e.g. Protter 24)); ad the third oe uses the fact that q 2, the properties of the fuctio φx) ad a similar argumet as i 4.1). For the first itegral o the right-had side of 4.13) if β > 1 we ca pick some q such that max{p, 1} q < β ad make exactly the same steps as i 4.14) to get E 1/β t p κx) µds, dx) K. 4.15) x >s 1/β Whe β = 1, ν is symmetric so the itegratio with respect to µ is the same as the itegratio with respect to the compesated measure µ aroud the origi. Thus, whe β = 1, for the first itegral o the right-had side of 4.13) we ca write E 1/β = p /β p /β E t p κx) µds, dx) x >s 1/β t t p E κx)µds, dx) κx)dsνdx) x >s 1/β x >s 1/β t ) κx) p νds, dx) + K K, 4.16) x >s 1/β where for the first iequality we made use of the fact that p 1. Fially, for the last itegral o the right-had side of 4.13), sice the jump measure has a bouded support i.e. there are o jumps higher tha ɛ i absolute value) the expectatio exists ad for the case β 1 we trivially have t p E κ x)µds, dx) K 1 p /β. 4.17) 1/β 21
22 Whe β < 1 recall κx) = i this case) we split t κ x)µds, dx) = t x s 1/β xµds, dx) + t x >s 1/β xµds, dx), ad use the fact that for arbitrary < m 1 ad some a ad b we have a + b m a m + b m, to prove E 1/β t p κ x)µds, dx) K. 4.18) Combiig 4.14)-4.18) we get 4.13), ad thus the poitwise covergece. To prove the covergece of p/β E Ψ t p ) for the local uiform topology o t we eed oly show that this sequece is relatively compact. For this we use the Ascoli-Arzela s Theorem see e.g. Jacod ad Shiryaev 23), Theorem VI.1.5). The relative compactess of p/β E Ψ t p ) follows from the fact that the bouds i 4.14)-4.17) are cotiuous i t. This proves part b) of the Lemma. Lemma 2 Cosider the probability space Ω, F, P) o which the process X t is defied via 4.1). Let u s, l s ad h s be some processes o this probability space which are adapted, bouded, ad have càdlàg paths. Assume that A2 holds for νx) ad that νx) is symmetric whe β = 1. Fially, if β < 1 set κx) ad whe β 1 let κx) be a trucatio fuctio, which is symmetric i the case β = 1. The, for arbitrary ɛ > ad some q 1 > β ad p q 2 < β we have E K p/β x ɛ E u s κx)1 {ls <y<h s } µds, dx, dy) + + )) p/q1 sup { u s q 1 h s l s } + K p/β s where the costat K does ot deped o. E u s κ x)1 {ls <y<h s }µds, dx, dy) x ɛ + )) p/q2 sup { u s q 2 h s l s }, s 4.19) Proof: The proof of the Lemma is very similar to the proof of Lemma 1, part b). First we make the decompositio x ɛ = + u s 1 {ls <y<h s }κx) µds, dx, dy) + s 1/β < x ɛ x s 1/β ɛ u s 1 {ls <y<h s }κx) µds, dx, dy) + + u s 1 {ls <y<h s }κx) µds, dx, dy). 4.2) We proceed with boudig the p-th absolute momet of the two itegrals o the right had side of the above equatio for the case β 1. For the last oe we choose some 1 q 2, such that p 22
23 q > β ad apply the Burkholder-Davis-Gudy iequality to get p E u s 1 {ls <y<h s }κx) µds, dx, dy) x s 1/β ɛ + q) p/q E u s 1 {ls <y<h s }κx) µds, dx, dy) x s 1/β ɛ + )) p/q K E u s q κx) q 1 {ls <y<h s }νx)dsdxdy x s 1/β + ) p/q K p/β E sup { u s q h s l s }, 4.21) s where the costat K is chagig from lie to lie. For the secod itegral i 4.2) we proceed similar to 4.15) ad 4.16) to get p E u s 1 {ls <y<h s }κx) µds, dx, dy) s 1/β < x ɛ + )) p/q K p/β E sup { u s q h s l s }, 4.22) s for some q such that max{p, 1} q < β if β > 1 ad q = p if β = 1. Similar trasformatios as i 4.17)-4.18) ad usig the fact that we itegrate over the bouded regio x ɛ give p E x ɛ K p/β E + u s 1 {ls <y<h s }κ x)µds, dx, dy) sup { u s p h s l s } s ) + K p/β E )) p/q sup { u s q h s l s }, s 4.23) where β < q < 1. Combiig iequalities 4.21)-4.23) we prove the Lemma. 4.2 Limit theorems for realized power variatio i pure-jump case Usig the prelimiary results we ow prove the followig theorem. Theorem 2 For the process X t defied i 2.2) assume that A1b), A2 ad A3 hold. Deote with L s a pure-jump Lévy process defied o some probability space) which has Lévy measure ds ν 1 x)dx ad drift { b = κ x)ν 1 x)dx if β 1 κx)ν 1x)dx if β < 1, with respect to the trucatio fuctio κ used i the defiitio of X t. If g p s) = E L s p ), the for p < β we have 1 p/β u.c.p. V p, X, ) t t g p a s ) σ 2s p ds. 4.1) Proof: We prove the theorem uder the followig stroger assumptio A1 : Assumptio A1. I additio to assumptio A1 assume that the processes a s, b 2s ad σ 2s are bouded. 23
24 To prove that if the claim i the theorem is true uder the stroger assumptio A1, it holds also uder assumptio A1, we ca use exactly the same localizatio argumet as i the proof of Lemma 4.6 i Jacod 28). Therefore we omit this part of the proof ad proceed with provig Theorem 1 uder the stroger assumptio A1. Also as for the lemmas i the previous subsectio we will use the equivalet i probability) represetatio of X give i 4.1). First, we eed some otatio. ecall that for X t we work with its represetatio i 4.1). For arbitrary ɛ > we set t Xɛ) t = X t σ 2s x1 {y<as }µds, dx, dy), X ɛ) t = X t Xɛ) t. 4.2) x >ɛ + Further, deote with S 1, S 2,..., S m,... the sequece of successive jump times of X ɛ) i.e. the jumps of the process X t which after scalig by σ 2Sm are higher tha ɛ i absolute value). Furthermore, if Ω T ) deotes the set of ω-s for which each iterval [, T ] i 1), i ] cotais at most oe S m ad i additio i Xɛ) Kɛ for some costat K recall the process σ 2t is bouded pathwise), the Ω T ) Ω as. Fially, for each S m, we set m = i Xɛ), where i 1), i ] is the iterval cotaiig S m. With this otatio o the set Ω T ) we have V p, X, ) t V p, Xɛ), ) t = { m + X Sm p m p } <S m t K S m t X Sm p + K, t T, 4.3) where K is some costat ad the sum i 4.3) is of course well defied ad almost surely fiite the iequality is trivial for p 1 ad for p > 1 it from a Taylor expasio). Therefore, sice Ω T ) Ω we have Thus, to fiish the proof we eed oly show 1 p/β {V p, X, ) t V p, Xɛ), ) t } u.c.p.. 1 p/β u.c.p. V p, Xɛ), ) t t g p a s ) σ 2s p ds. 4.4) We make a slight abuse of otatio for κx), as i the proof of Lemma 2 part b), i order to aalyze the differet cases for β together. I particular, we set κx) whe β < 1, which correspods to o compesatio of the small jumps, ad set b 2t see assumptio A3). We start the proof of 4.4) by makig the followig decompositio. 1 p/β V p, Xɛ), ) t t A 1 = 1 p/β g p a s ) σ 2s p ds = A 1 + A 2 + A 3, 4.5) [t/ ] i=1 ξ i, ξ i = { ξ i1 p ξ i2 p }, ξ i1 = i Xɛ), ξ i2 = i i 1) i + i 1) x ɛ x ɛ σ 2i 1) 1 {y<ai 1) }κx) µds, dx, dy) + σ 2i 1) 1 {y<ai 1) }κ x)µds, dx, dy) + 24
25 [t/ ] A 3 = + [t/ ] A 2 = g p a i 1) ) σ 2i 1) p i=1 i=1 t g p a s ) σ 2s p ds, { i σ 2i 1) p p/β 1 {y<ai 1) }κx) µds, dx, dy) i 1) x ɛ + p } 1 {y<ai 1) }κ x)µds, dx, dy) g p a i 1) ). x ɛ + We will show that each of the three terms o the right-had side of 4.5) coverges i probability, uiformly i time, to. We start with A 1. I what follows E i 1 ) is shorthad for E F ) i 1). We wat to derive boud for E i 1 ξi. If p 1 we ca make use of the basic iequality a + b p a p + b p for arbitrary a ad b. If p > 1, we ca use first-order Taylor expasio ad the apply the Hölder s iequality. Both cases lead to where E i 1 ξ i K E i 1 ζi p) 1 p 1 E i 1 ξi1 p) p 1 p + K E i 1 ζi p) 1 p 1 E i 1 ξi2 p) p 1 p, ζ i = i i 1) i + i 1) x ɛ x ɛ σ 2s 1 {y<as } σ 2i 1) 1 {y<ai 1) })κx) µds, dx, dy) + σ 2s 1 {y<as } σ 2i 1) 1 {y<ai 1) })κ x)µds, dx, dy). + Usig Lemma 2 ad the fact that the processes σ 2s ad a s are bouded, we have E i 1 ξ i K p 1 β To boud E i 1 ζ i p we first use the followig decompositio E i 1 ζ i p) 1 p ) σ 2s 1 {y<as } σ 2i 1) 1 {y<ai 1) } = σ 2s 1 {y<as } 1 {y<ai 1) } )+σ 2s σ 2i 1) )1 {y<ai 1) }, ad the apply Lemma 2 to get E i 1 ζ i p K p/β E i 1 sup i 1) s i σ 2s σ 2i 1) p +K p/β E i 1 sup a s a i 1) i 1) s i +K p/β +K p/β E i 1 E i 1 ) p/q sup σ 2s σ 2i 1) q i 1) s i ) p/q sup a s a i 1), i 1) s i where q is some umber higher or at most equal to p. But [t/ ) ] E sup σ 2s σ 2i 1) r i 1) s i = t E i=1 sup σ 2u σ 2[u/ ] r [u/ ] u [u/ ] + 25 ) du,
26 for some arbitrary r. The for arbitrary u > we have lim E sup σ 2u σ 2[u/ ] r [u/ ] u [u/ ] + ) CE σ u r ). Therefore, usig the Lebesque s covergece theorem ad the fact that σ 2s is a bouded càdlàg process uder A1 ), we have [t/ ) ] E sup σ 2s σ 2i 1) r. i 1) s i i=1 Similar result holds for a s. Therefore we have [t/ ] i=1 E i 1 ξ i, u.c.p. ad thus A 1. For A 2 usig iema itegrability we have that A 2 coverges poitwise i ω ad locally uiformly i time to. Fially, for A 3 first ote that i i 1) x ɛ + 1 {y<ai 1) } κx) µds, dx, dy)+ i i 1) x ɛ + 1 {y<ai 1) } κ x)µds, dx, dy), is equal i distributio to L a i 1), where L is the Lévy process defied i Lemma 1 it is defied o some possibly) differet probability space from the origial oe o which X t is defied, but the choice of the costat ɛ is of course the same). The we ca use the covergece result i 4.3), which recall is locally uiform i time. Therefore, sice a s is bouded, we have i E i 1 p/β 1 {y<ai 1) }κx) µds, dx, dy) i 1) x ɛ + i p + 1 {y<ai 1) }κ x)µds, dx, dy) g p a i 1) ) i 1) x ɛ + f ), where f ) is some fuctio such that lim f ) = ad therefore A 3 u.c.p Proof of Theorem 1 i the paper We prove all three parts together. O a set Ω Ω, β,t ] Υ, p) is a cotiuous trasformatio of V p, Υ, k ) T ad V p, Υ, ) T. Therefore we will be doe if we ca show that for a fixed T > ad some p l < p u which for part b) ad c) of the Theorem are such that β is outside [p l, p u ]) we have Π p) := 1 p/β Υ,T V p, Υ, k ) T 1 p/β Υ,T V p, Υ, ) T ) ) k p/β Υ,T 1 C T p) C T p) P ) uiformly o [p l, p u ], 4.1) where C T p) is oe of the limits i 3.3) ad 3.5) depedig o whether Υ Y, Z or Υ X respectively ad the covergece i 4.1) is i C[p l, p u ], 2 +) the space of 2 +-valued cotiuous fuctios o [p l, p u ]) equipped with the uiform metric. The proof of 4.1) cosists of establishig fiite-dimesioal covergece ad tightess of the sequece o the left-had side of 4.1). The 26
27 fiite-dimesioal covergece is trivial. It follows from the poitwise i p) covergece results of Sectio 3.1 i the paper. Therefore we eed oly prove tightess. To this ed, deote the modulus of cotiuity o C[p l, p u ], 2 +) see e.g. Jacod ad Shiryaev 23), VI.1) Set wπ, θ) := sup{ sup u,v [p,p+θ] [T/ ] UΥ, p, q, ) T = i=1 With this otatio for sufficietly small θ we have Π u) Π v) : p l p p + θ p u }. 1/β Υ,T i Υ p 1/β Υ,T i Υ q. 4.2) wπ, θ) UΥ, p l + θ, p l, ) T + UΥ, p u, p u θ, ) T +UΥ, p l + θ, p l, k ) T + UΥ, p u, p u θ, k ) T. 4.3) The we ca use the iequality [T/ ] UΥ, p, q, ) T K p q i=1 1/β Υ,T i Υ p q ɛ + 1/β Υ,T i Υ p q+ɛ), 4.4) for arbitrary small ɛ ad a costat K. Therefore the Ascoli-Arzela s criteria for tightess is satisfied see Jacod ad Shiryaev 23), VI.3.26ii)) ad together with the fiite-dimesioal covergece, this implies 4.1) ad hece the result of the Theorem. 4.4 Proof of Theorem 2 i the paper Part b) of the theorem follows from Bardorff-Nielse ad Shephard 24, 26) ad Bardorff- Nielse et al. 26), hece we show oly part a). We proof the result i 3.13) uder the stroger assumptio A1, the extesio to the weaker assumptio A1 follows by a localisatio argumet. Usig the fact that p 1 we have trivially V p, Z, ) t V p, Y, ) t V p, X, ) t. 4.1) At the same time a straightforward applicatio of Jese s iequality yields Combiig 4.1) ad 4.2) we have p BG T X) ε ) 1 E i 1 i X p K for ay ε >. 4.2) 1/2 p/2 V p, Z, ) t V p, Y, ) t u.c.p., 4.3) uder the coditios of the theorem. Therefore it is eough to prove 3.13) for Υ Y. But this trivially follows from the followig applicatio of Theorem 7.3 i Jacod 27) ) 1 k ) 1 p/2 t V p, Y, k ) t µ p σ 1u p du 1 p/2 t V p, Y, ) t µ p σ 1u p du L s Ξ t, 4.4) where the process Ξ t is defied o a extesio of the origial probability space, is cotiuous, ad coditioally o the σ-field F is cetered Gaussia with variace-covariace matrix give by t σ 1u 2 pdu kµ 2p µ 2 p) k 1 p/2 µk, p) kµ 2 p k 1 p/2 µ p k) kµ 2 p µ 2p µ 2 p The the covergece i 3.13) follows trivially from the result i 4.4) ad a Delta method ote that the covergece i 4.3) is stable). ). 27
28 efereces Bardorff-Nielse, O. ad N. Shephard 24). Power ad Bipower Variatio with Stochastic Volatility ad Jumps. Joural of Fiacial Ecoometrics 2, Bardorff-Nielse, O. ad N. Shephard 26). Ecoometrics of Testig for Jumps i Fiacial Ecoomics usig Bipower Variatio. Joural of Fiacial Ecoometrics 4, 1 3. Bardorff-Nielse, O. ad N. Shephard 27). Variatio, jumps, Market Frictios ad High Frequecy Data i Facial Ecoometrics. I. Bludell, T. Persso, ad W. Newey Eds.), Advaces i Ecoomics ad Ecoometrics. Theory ad Applicatios, Nith World Cogress. Cambridge Uiversity Press. Bardorff-Nielse, O., N. Shephard, ad M. Wikel 26). Limit Theorems for Multipower Variatio i the Presece of Jumps i Fiacial Ecoometrics. Stochastic Processes ad Their Applicatios 116, Jacod, J. 27). Statistics ad High-Frequecy Data. Semstat Course i La Maga. Jacod, J. 28). Asymptotic Properties of Power Variatios ad Associated Fuctioals of Semimartigales. Stochastic Processes ad their Applicatios 118, Jacod, J. ad A. N. Shiryaev 23). Limit Theorems For Stochastic Processes 2d ed.). Berli: Spriger-Verlag. Protter, P. 24). Stochastic Itegratio ad Differetial Equatios 2d ed.). Berli: Spriger- Verlag. Todorov, V. ad G. Tauche 29). Activity Sigature Fuctios for High-Frequecy Data Aalysis. Workig paper, Duke Uiversity ad Northwester Uiversity. Woerer, J. 26). Aalyzig the Fie Structure of Cotiuous-Time Stochastic Processes. Workig paper. 28
Kolmogorov-Smirnov type Tests for Local Gaussianity in High-Frequency Data
Proceedigs 59th ISI World Statistics Cogress, 5-30 August 013, Hog Kog (Sessio STS046) p.09 Kolmogorov-Smirov type Tests for Local Gaussiaity i High-Frequecy Data George Tauche, Duke Uiversity Viktor Todorov,
More informationChapter 3. Strong convergence. 3.1 Definition of almost sure convergence
Chapter 3 Strog covergece As poited out i the Chapter 2, there are multiple ways to defie the otio of covergece of a sequece of radom variables. That chapter defied covergece i probability, covergece i
More informationActivity Signature Functions for High-Frequency Data Analysis
Activity Sigature Fuctios for High-Frequecy Data Aalysis Viktor Todorov ad George Tauche This Draft: July, 8 First Draft: November 7 Abstract We defie a ew cocept termed the activity sigature fuctio, which
More informationAn alternative proof of a theorem of Aldous. concerning convergence in distribution for martingales.
A alterative proof of a theorem of Aldous cocerig covergece i distributio for martigales. Maurizio Pratelli Dipartimeto di Matematica, Uiversità di Pisa. Via Buoarroti 2. I-56127 Pisa, Italy e-mail: pratelli@dm.uipi.it
More informationSUPPLEMENT TO REALIZED LAPLACE TRANSFORMS FOR PURE-JUMP SEMIMARTINGALES
SUPPLMN O ALIZD LAPLAC ANSFOMS FO PU-JUMP SMIMAINGALS By Viktor odorov ad George auche Northwester Uiversity ad Duke Uiversity his supplemet cotais some techical results ad proofs of theorems i the paper.
More informationLecture Chapter 6: Convergence of Random Sequences
ECE5: Aalysis of Radom Sigals Fall 6 Lecture Chapter 6: Covergece of Radom Sequeces Dr Salim El Rouayheb Scribe: Abhay Ashutosh Doel, Qibo Zhag, Peiwe Tia, Pegzhe Wag, Lu Liu Radom sequece Defiitio A ifiite
More informationLecture 19: Convergence
Lecture 19: Covergece Asymptotic approach I statistical aalysis or iferece, a key to the success of fidig a good procedure is beig able to fid some momets ad/or distributios of various statistics. I may
More informationSingular Continuous Measures by Michael Pejic 5/14/10
Sigular Cotiuous Measures by Michael Peic 5/4/0 Prelimiaries Give a set X, a σ-algebra o X is a collectio of subsets of X that cotais X ad ad is closed uder complemetatio ad coutable uios hece, coutable
More information6.3 Testing Series With Positive Terms
6.3. TESTING SERIES WITH POSITIVE TERMS 307 6.3 Testig Series With Positive Terms 6.3. Review of what is kow up to ow I theory, testig a series a i for covergece amouts to fidig the i= sequece of partial
More information7.1 Convergence of sequences of random variables
Chapter 7 Limit Theorems Throughout this sectio we will assume a probability space (, F, P), i which is defied a ifiite sequece of radom variables (X ) ad a radom variable X. The fact that for every ifiite
More informationResampling Methods. X (1/2), i.e., Pr (X i m) = 1/2. We order the data: X (1) X (2) X (n). Define the sample median: ( n.
Jauary 1, 2019 Resamplig Methods Motivatio We have so may estimators with the property θ θ d N 0, σ 2 We ca also write θ a N θ, σ 2 /, where a meas approximately distributed as Oce we have a cosistet estimator
More informationMATH 320: Probability and Statistics 9. Estimation and Testing of Parameters. Readings: Pruim, Chapter 4
MATH 30: Probability ad Statistics 9. Estimatio ad Testig of Parameters Estimatio ad Testig of Parameters We have bee dealig situatios i which we have full kowledge of the distributio of a radom variable.
More informationFall 2013 MTH431/531 Real analysis Section Notes
Fall 013 MTH431/531 Real aalysis Sectio 8.1-8. Notes Yi Su 013.11.1 1. Defiitio of uiform covergece. We look at a sequece of fuctios f (x) ad study the coverget property. Notice we have two parameters
More informationDiscrete Mathematics for CS Spring 2008 David Wagner Note 22
CS 70 Discrete Mathematics for CS Sprig 2008 David Wager Note 22 I.I.D. Radom Variables Estimatig the bias of a coi Questio: We wat to estimate the proportio p of Democrats i the US populatio, by takig
More informationConvergence of random variables. (telegram style notes) P.J.C. Spreij
Covergece of radom variables (telegram style otes).j.c. Spreij this versio: September 6, 2005 Itroductio As we kow, radom variables are by defiitio measurable fuctios o some uderlyig measurable space
More informationSequences and Series of Functions
Chapter 6 Sequeces ad Series of Fuctios 6.1. Covergece of a Sequece of Fuctios Poitwise Covergece. Defiitio 6.1. Let, for each N, fuctio f : A R be defied. If, for each x A, the sequece (f (x)) coverges
More informationIntroduction to Extreme Value Theory Laurens de Haan, ISM Japan, Erasmus University Rotterdam, NL University of Lisbon, PT
Itroductio to Extreme Value Theory Laures de Haa, ISM Japa, 202 Itroductio to Extreme Value Theory Laures de Haa Erasmus Uiversity Rotterdam, NL Uiversity of Lisbo, PT Itroductio to Extreme Value Theory
More information7.1 Convergence of sequences of random variables
Chapter 7 Limit theorems Throughout this sectio we will assume a probability space (Ω, F, P), i which is defied a ifiite sequece of radom variables (X ) ad a radom variable X. The fact that for every ifiite
More informationRandom Variables, Sampling and Estimation
Chapter 1 Radom Variables, Samplig ad Estimatio 1.1 Itroductio This chapter will cover the most importat basic statistical theory you eed i order to uderstad the ecoometric material that will be comig
More informationProduct measures, Tonelli s and Fubini s theorems For use in MAT3400/4400, autumn 2014 Nadia S. Larsen. Version of 13 October 2014.
Product measures, Toelli s ad Fubii s theorems For use i MAT3400/4400, autum 2014 Nadia S. Larse Versio of 13 October 2014. 1. Costructio of the product measure The purpose of these otes is to preset the
More informationEmpirical Processes: Glivenko Cantelli Theorems
Empirical Processes: Gliveko Catelli Theorems Mouliath Baerjee Jue 6, 200 Gliveko Catelli classes of fuctios The reader is referred to Chapter.6 of Weller s Torgo otes, Chapter??? of VDVW ad Chapter 8.3
More informationInfinite Sequences and Series
Chapter 6 Ifiite Sequeces ad Series 6.1 Ifiite Sequeces 6.1.1 Elemetary Cocepts Simply speakig, a sequece is a ordered list of umbers writte: {a 1, a 2, a 3,...a, a +1,...} where the elemets a i represet
More informationNotes 19 : Martingale CLT
Notes 9 : Martigale CLT Math 733-734: Theory of Probability Lecturer: Sebastie Roch Refereces: [Bil95, Chapter 35], [Roc, Chapter 3]. Sice we have ot ecoutered weak covergece i some time, we first recall
More informationDistribution of Random Samples & Limit theorems
STAT/MATH 395 A - PROBABILITY II UW Witer Quarter 2017 Néhémy Lim Distributio of Radom Samples & Limit theorems 1 Distributio of i.i.d. Samples Motivatig example. Assume that the goal of a study is to
More informationDetailed proofs of Propositions 3.1 and 3.2
Detailed proofs of Propositios 3. ad 3. Proof of Propositio 3. NB: itegratio sets are geerally omitted for itegrals defied over a uit hypercube [0, s with ay s d. We first give four lemmas. The proof of
More informationLet us give one more example of MLE. Example 3. The uniform distribution U[0, θ] on the interval [0, θ] has p.d.f.
Lecture 5 Let us give oe more example of MLE. Example 3. The uiform distributio U[0, ] o the iterval [0, ] has p.d.f. { 1 f(x =, 0 x, 0, otherwise The likelihood fuctio ϕ( = f(x i = 1 I(X 1,..., X [0,
More informationParameter, Statistic and Random Samples
Parameter, Statistic ad Radom Samples A parameter is a umber that describes the populatio. It is a fixed umber, but i practice we do ot kow its value. A statistic is a fuctio of the sample data, i.e.,
More informationJournal of Multivariate Analysis. Superefficient estimation of the marginals by exploiting knowledge on the copula
Joural of Multivariate Aalysis 102 (2011) 1315 1319 Cotets lists available at ScieceDirect Joural of Multivariate Aalysis joural homepage: www.elsevier.com/locate/jmva Superefficiet estimatio of the margials
More informationMASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.265/15.070J Fall 2013 Lecture 21 11/27/2013
MASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.265/15.070J Fall 2013 Lecture 21 11/27/2013 Fuctioal Law of Large Numbers. Costructio of the Wieer Measure Cotet. 1. Additioal techical results o weak covergece
More informationStochastic Simulation
Stochastic Simulatio 1 Itroductio Readig Assigmet: Read Chapter 1 of text. We shall itroduce may of the key issues to be discussed i this course via a couple of model problems. Model Problem 1 (Jackso
More informationMASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.436J/15.085J Fall 2008 Lecture 19 11/17/2008 LAWS OF LARGE NUMBERS II THE STRONG LAW OF LARGE NUMBERS
MASSACHUSTTS INSTITUT OF TCHNOLOGY 6.436J/5.085J Fall 2008 Lecture 9 /7/2008 LAWS OF LARG NUMBRS II Cotets. The strog law of large umbers 2. The Cheroff boud TH STRONG LAW OF LARG NUMBRS While the weak
More informationChapter 6 Infinite Series
Chapter 6 Ifiite Series I the previous chapter we cosidered itegrals which were improper i the sese that the iterval of itegratio was ubouded. I this chapter we are goig to discuss a topic which is somewhat
More informationDS 100: Principles and Techniques of Data Science Date: April 13, Discussion #10
DS 00: Priciples ad Techiques of Data Sciece Date: April 3, 208 Name: Hypothesis Testig Discussio #0. Defie these terms below as they relate to hypothesis testig. a) Data Geeratio Model: Solutio: A set
More informationLimit Theorems for the Empirical Distribution Function of Scaled Increments of Itô Semimartingales at high frequencies
Limit Theorems for the Empirical Distributio Fuctio of Scaled Icremets of Itô Semimartigales at high frequecies Viktor Todorov ad George Tauche February 4, 03 Abstract We derive limit theorems for the
More informationProperties and Hypothesis Testing
Chapter 3 Properties ad Hypothesis Testig 3.1 Types of data The regressio techiques developed i previous chapters ca be applied to three differet kids of data. 1. Cross-sectioal data. 2. Time series data.
More informationSolutions to HW Assignment 1
Solutios to HW: 1 Course: Theory of Probability II Page: 1 of 6 Uiversity of Texas at Austi Solutios to HW Assigmet 1 Problem 1.1. Let Ω, F, {F } 0, P) be a filtered probability space ad T a stoppig time.
More informationEcon 325/327 Notes on Sample Mean, Sample Proportion, Central Limit Theorem, Chi-square Distribution, Student s t distribution 1.
Eco 325/327 Notes o Sample Mea, Sample Proportio, Cetral Limit Theorem, Chi-square Distributio, Studet s t distributio 1 Sample Mea By Hiro Kasahara We cosider a radom sample from a populatio. Defiitio
More informationLecture 2: Monte Carlo Simulation
STAT/Q SCI 43: Itroductio to Resamplig ethods Sprig 27 Istructor: Ye-Chi Che Lecture 2: ote Carlo Simulatio 2 ote Carlo Itegratio Assume we wat to evaluate the followig itegratio: e x3 dx What ca we do?
More information(A sequence also can be thought of as the list of function values attained for a function f :ℵ X, where f (n) = x n for n 1.) x 1 x N +k x N +4 x 3
MATH 337 Sequeces Dr. Neal, WKU Let X be a metric space with distace fuctio d. We shall defie the geeral cocept of sequece ad limit i a metric space, the apply the results i particular to some special
More informationIntegrable Functions. { f n } is called a determining sequence for f. If f is integrable with respect to, then f d does exist as a finite real number
MATH 532 Itegrable Fuctios Dr. Neal, WKU We ow shall defie what it meas for a measurable fuctio to be itegrable, show that all itegral properties of simple fuctios still hold, ad the give some coditios
More informationCEE 522 Autumn Uncertainty Concepts for Geotechnical Engineering
CEE 5 Autum 005 Ucertaity Cocepts for Geotechical Egieerig Basic Termiology Set A set is a collectio of (mutually exclusive) objects or evets. The sample space is the (collectively exhaustive) collectio
More information4. Partial Sums and the Central Limit Theorem
1 of 10 7/16/2009 6:05 AM Virtual Laboratories > 6. Radom Samples > 1 2 3 4 5 6 7 4. Partial Sums ad the Cetral Limit Theorem The cetral limit theorem ad the law of large umbers are the two fudametal theorems
More informationSequences. Notation. Convergence of a Sequence
Sequeces A sequece is essetially just a list. Defiitio (Sequece of Real Numbers). A sequece of real umbers is a fuctio Z (, ) R for some real umber. Do t let the descriptio of the domai cofuse you; it
More informationA Weak Law of Large Numbers Under Weak Mixing
A Weak Law of Large Numbers Uder Weak Mixig Bruce E. Hase Uiversity of Wiscosi Jauary 209 Abstract This paper presets a ew weak law of large umbers (WLLN) for heterogeous depedet processes ad arrays. The
More information1 of 7 7/16/2009 6:06 AM Virtual Laboratories > 6. Radom Samples > 1 2 3 4 5 6 7 6. Order Statistics Defiitios Suppose agai that we have a basic radom experimet, ad that X is a real-valued radom variable
More information32 estimating the cumulative distribution function
32 estimatig the cumulative distributio fuctio 4.6 types of cofidece itervals/bads Let F be a class of distributio fuctios F ad let θ be some quatity of iterest, such as the mea of F or the whole fuctio
More informationKernel density estimator
Jauary, 07 NONPARAMETRIC ERNEL DENSITY ESTIMATION I this lecture, we discuss kerel estimatio of probability desity fuctios PDF Noparametric desity estimatio is oe of the cetral problems i statistics I
More informationTrading Friction Noise 1
Ecoomics 883 Sprig 205 Tauche Tradig Frictio Noise Setup Let Y be the usual cotiuous semi-martigale Y t = t 0 cs dw s We will cosider jump discotiuities later. The usual setup for modelig tradig frictio
More informationAnalytic Continuation
Aalytic Cotiuatio The stadard example of this is give by Example Let h (z) = 1 + z + z 2 + z 3 +... kow to coverge oly for z < 1. I fact h (z) = 1/ (1 z) for such z. Yet H (z) = 1/ (1 z) is defied for
More informationMath 341 Lecture #31 6.5: Power Series
Math 341 Lecture #31 6.5: Power Series We ow tur our attetio to a particular kid of series of fuctios, amely, power series, f(x = a x = a 0 + a 1 x + a 2 x 2 + where a R for all N. I terms of a series
More informationBerry-Esseen bounds for self-normalized martingales
Berry-Essee bouds for self-ormalized martigales Xiequa Fa a, Qi-Ma Shao b a Ceter for Applied Mathematics, Tiaji Uiversity, Tiaji 30007, Chia b Departmet of Statistics, The Chiese Uiversity of Hog Kog,
More informationOutput Analysis and Run-Length Control
IEOR E4703: Mote Carlo Simulatio Columbia Uiversity c 2017 by Marti Haugh Output Aalysis ad Ru-Legth Cotrol I these otes we describe how the Cetral Limit Theorem ca be used to costruct approximate (1 α%
More information1 Convergence in Probability and the Weak Law of Large Numbers
36-752 Advaced Probability Overview Sprig 2018 8. Covergece Cocepts: i Probability, i L p ad Almost Surely Istructor: Alessadro Rialdo Associated readig: Sec 2.4, 2.5, ad 4.11 of Ash ad Doléas-Dade; Sec
More informationThe natural exponential function
The atural expoetial fuctio Attila Máté Brookly College of the City Uiversity of New York December, 205 Cotets The atural expoetial fuctio for real x. Beroulli s iequality.....................................2
More informationSTAT331. Example of Martingale CLT with Cox s Model
STAT33 Example of Martigale CLT with Cox s Model I this uit we illustrate the Martigale Cetral Limit Theorem by applyig it to the partial likelihood score fuctio from Cox s model. For simplicity of presetatio
More informationMAT1026 Calculus II Basic Convergence Tests for Series
MAT026 Calculus II Basic Covergece Tests for Series Egi MERMUT 202.03.08 Dokuz Eylül Uiversity Faculty of Sciece Departmet of Mathematics İzmir/TURKEY Cotets Mootoe Covergece Theorem 2 2 Series of Real
More informationSeunghee Ye Ma 8: Week 5 Oct 28
Week 5 Summary I Sectio, we go over the Mea Value Theorem ad its applicatios. I Sectio 2, we will recap what we have covered so far this term. Topics Page Mea Value Theorem. Applicatios of the Mea Value
More informationMASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.265/15.070J Fall 2013 Lecture 2 9/9/2013. Large Deviations for i.i.d. Random Variables
MASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.265/15.070J Fall 2013 Lecture 2 9/9/2013 Large Deviatios for i.i.d. Radom Variables Cotet. Cheroff boud usig expoetial momet geeratig fuctios. Properties of a momet
More informationAn Introduction to Asymptotic Theory
A Itroductio to Asymptotic Theory Pig Yu School of Ecoomics ad Fiace The Uiversity of Hog Kog Pig Yu (HKU) Asymptotic Theory 1 / 20 Five Weapos i Asymptotic Theory Five Weapos i Asymptotic Theory Pig Yu
More informationRates of Convergence by Moduli of Continuity
Rates of Covergece by Moduli of Cotiuity Joh Duchi: Notes for Statistics 300b March, 017 1 Itroductio I this ote, we give a presetatio showig the importace, ad relatioship betwee, the modulis of cotiuity
More informationNotes 5 : More on the a.s. convergence of sums
Notes 5 : More o the a.s. covergece of sums Math 733-734: Theory of Probability Lecturer: Sebastie Roch Refereces: Dur0, Sectios.5; Wil9, Sectio 4.7, Shi96, Sectio IV.4, Dur0, Sectio.. Radom series. Three-series
More informationStatistics 511 Additional Materials
Cofidece Itervals o mu Statistics 511 Additioal Materials This topic officially moves us from probability to statistics. We begi to discuss makig ifereces about the populatio. Oe way to differetiate probability
More informationAdvanced Stochastic Processes.
Advaced Stochastic Processes. David Gamarik LECTURE 2 Radom variables ad measurable fuctios. Strog Law of Large Numbers (SLLN). Scary stuff cotiued... Outlie of Lecture Radom variables ad measurable fuctios.
More informationFrequentist Inference
Frequetist Iferece The topics of the ext three sectios are useful applicatios of the Cetral Limit Theorem. Without kowig aythig about the uderlyig distributio of a sequece of radom variables {X i }, for
More informationMeasure and Measurable Functions
3 Measure ad Measurable Fuctios 3.1 Measure o a Arbitrary σ-algebra Recall from Chapter 2 that the set M of all Lebesgue measurable sets has the followig properties: R M, E M implies E c M, E M for N implies
More informationIntroductory statistics
CM9S: Machie Learig for Bioiformatics Lecture - 03/3/06 Itroductory statistics Lecturer: Sriram Sakararama Scribe: Sriram Sakararama We will provide a overview of statistical iferece focussig o the key
More informationAppendix to: Hypothesis Testing for Multiple Mean and Correlation Curves with Functional Data
Appedix to: Hypothesis Testig for Multiple Mea ad Correlatio Curves with Fuctioal Data Ao Yua 1, Hog-Bi Fag 1, Haiou Li 1, Coli O. Wu, Mig T. Ta 1, 1 Departmet of Biostatistics, Bioiformatics ad Biomathematics,
More informationEcon 325 Notes on Point Estimator and Confidence Interval 1 By Hiro Kasahara
Poit Estimator Eco 325 Notes o Poit Estimator ad Cofidece Iterval 1 By Hiro Kasahara Parameter, Estimator, ad Estimate The ormal probability desity fuctio is fully characterized by two costats: populatio
More informationSolution. 1 Solutions of Homework 1. Sangchul Lee. October 27, Problem 1.1
Solutio Sagchul Lee October 7, 017 1 Solutios of Homework 1 Problem 1.1 Let Ω,F,P) be a probability space. Show that if {A : N} F such that A := lim A exists, the PA) = lim PA ). Proof. Usig the cotiuity
More informationA statistical method to determine sample size to estimate characteristic value of soil parameters
A statistical method to determie sample size to estimate characteristic value of soil parameters Y. Hojo, B. Setiawa 2 ad M. Suzuki 3 Abstract Sample size is a importat factor to be cosidered i determiig
More informationMA131 - Analysis 1. Workbook 2 Sequences I
MA3 - Aalysis Workbook 2 Sequeces I Autum 203 Cotets 2 Sequeces I 2. Itroductio.............................. 2.2 Icreasig ad Decreasig Sequeces................ 2 2.3 Bouded Sequeces..........................
More informationWeb-based Supplementary Materials for A Modified Partial Likelihood Score Method for Cox Regression with Covariate Error Under the Internal
Web-based Supplemetary Materials for A Modified Partial Likelihood Score Method for Cox Regressio with Covariate Error Uder the Iteral Validatio Desig by David M. Zucker, Xi Zhou, Xiaomei Liao, Yi Li,
More informationNotes 27 : Brownian motion: path properties
Notes 27 : Browia motio: path properties Math 733-734: Theory of Probability Lecturer: Sebastie Roch Refereces:[Dur10, Sectio 8.1], [MP10, Sectio 1.1, 1.2, 1.3]. Recall: DEF 27.1 (Covariace) Let X = (X
More information5. Likelihood Ratio Tests
1 of 5 7/29/2009 3:16 PM Virtual Laboratories > 9. Hy pothesis Testig > 1 2 3 4 5 6 7 5. Likelihood Ratio Tests Prelimiaries As usual, our startig poit is a radom experimet with a uderlyig sample space,
More informationSequences I. Chapter Introduction
Chapter 2 Sequeces I 2. Itroductio A sequece is a list of umbers i a defiite order so that we kow which umber is i the first place, which umber is i the secod place ad, for ay atural umber, we kow which
More informationStatistics and high frequency data
Statistics ad high frequecy data Jea Jacod Itroductio This short course is devoted to a few statistical problems related to the observatio of a give process o a fixed time iterval, whe the observatios
More informationLesson 10: Limits and Continuity
www.scimsacademy.com Lesso 10: Limits ad Cotiuity SCIMS Academy 1 Limit of a fuctio The cocept of limit of a fuctio is cetral to all other cocepts i calculus (like cotiuity, derivative, defiite itegrals
More informationOptimally Sparse SVMs
A. Proof of Lemma 3. We here prove a lower boud o the umber of support vectors to achieve geeralizatio bouds of the form which we cosider. Importatly, this result holds ot oly for liear classifiers, but
More informationMASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.265/15.070J Fall 2013 Lecture 3 9/11/2013. Large deviations Theory. Cramér s Theorem
MASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.265/5.070J Fall 203 Lecture 3 9//203 Large deviatios Theory. Cramér s Theorem Cotet.. Cramér s Theorem. 2. Rate fuctio ad properties. 3. Chage of measure techique.
More information3. Z Transform. Recall that the Fourier transform (FT) of a DT signal xn [ ] is ( ) [ ] = In order for the FT to exist in the finite magnitude sense,
3. Z Trasform Referece: Etire Chapter 3 of text. Recall that the Fourier trasform (FT) of a DT sigal x [ ] is ω ( ) [ ] X e = j jω k = xe I order for the FT to exist i the fiite magitude sese, S = x [
More informationLECTURE 8: ASYMPTOTICS I
LECTURE 8: ASYMPTOTICS I We are iterested i the properties of estimators as. Cosider a sequece of radom variables {, X 1}. N. M. Kiefer, Corell Uiversity, Ecoomics 60 1 Defiitio: (Weak covergece) A sequece
More informationLecture Notes for Analysis Class
Lecture Notes for Aalysis Class Topological Spaces A topology for a set X is a collectio T of subsets of X such that: (a) X ad the empty set are i T (b) Uios of elemets of T are i T (c) Fiite itersectios
More information11 THE GMM ESTIMATION
Cotets THE GMM ESTIMATION 2. Cosistecy ad Asymptotic Normality..................... 3.2 Regularity Coditios ad Idetificatio..................... 4.3 The GMM Iterpretatio of the OLS Estimatio.................
More informationThe Wasserstein distances
The Wasserstei distaces March 20, 2011 This documet presets the proof of the mai results we proved o Wasserstei distaces themselves (ad ot o curves i the Wasserstei space). I particular, triagle iequality
More informationCHAPTER 10 INFINITE SEQUENCES AND SERIES
CHAPTER 10 INFINITE SEQUENCES AND SERIES 10.1 Sequeces 10.2 Ifiite Series 10.3 The Itegral Tests 10.4 Compariso Tests 10.5 The Ratio ad Root Tests 10.6 Alteratig Series: Absolute ad Coditioal Covergece
More informationDefinition 4.2. (a) A sequence {x n } in a Banach space X is a basis for X if. unique scalars a n (x) such that x = n. a n (x) x n. (4.
4. BASES I BAACH SPACES 39 4. BASES I BAACH SPACES Sice a Baach space X is a vector space, it must possess a Hamel, or vector space, basis, i.e., a subset {x γ } γ Γ whose fiite liear spa is all of X ad
More informationRegression with an Evaporating Logarithmic Trend
Regressio with a Evaporatig Logarithmic Tred Peter C. B. Phillips Cowles Foudatio, Yale Uiversity, Uiversity of Aucklad & Uiversity of York ad Yixiao Su Departmet of Ecoomics Yale Uiversity October 5,
More informationRandom Walks on Discrete and Continuous Circles. by Jeffrey S. Rosenthal School of Mathematics, University of Minnesota, Minneapolis, MN, U.S.A.
Radom Walks o Discrete ad Cotiuous Circles by Jeffrey S. Rosethal School of Mathematics, Uiversity of Miesota, Mieapolis, MN, U.S.A. 55455 (Appeared i Joural of Applied Probability 30 (1993), 780 789.)
More informationAn Introduction to Randomized Algorithms
A Itroductio to Radomized Algorithms The focus of this lecture is to study a radomized algorithm for quick sort, aalyze it usig probabilistic recurrece relatios, ad also provide more geeral tools for aalysis
More informationChapter 7 Isoperimetric problem
Chapter 7 Isoperimetric problem Recall that the isoperimetric problem (see the itroductio its coectio with ido s proble) is oe of the most classical problem of a shape optimizatio. It ca be formulated
More informationThe random version of Dvoretzky s theorem in l n
The radom versio of Dvoretzky s theorem i l Gideo Schechtma Abstract We show that with high probability a sectio of the l ball of dimesio k cε log c > 0 a uiversal costat) is ε close to a multiple of the
More informationBasics of Probability Theory (for Theory of Computation courses)
Basics of Probability Theory (for Theory of Computatio courses) Oded Goldreich Departmet of Computer Sciece Weizma Istitute of Sciece Rehovot, Israel. oded.goldreich@weizma.ac.il November 24, 2008 Preface.
More informationBIRKHOFF ERGODIC THEOREM
BIRKHOFF ERGODIC THEOREM Abstract. We will give a proof of the poitwise ergodic theorem, which was first proved by Birkhoff. May improvemets have bee made sice Birkhoff s orgial proof. The versio we give
More information17. Joint distributions of extreme order statistics Lehmann 5.1; Ferguson 15
17. Joit distributios of extreme order statistics Lehma 5.1; Ferguso 15 I Example 10., we derived the asymptotic distributio of the maximum from a radom sample from a uiform distributio. We did this usig
More informationEntropy Rates and Asymptotic Equipartition
Chapter 29 Etropy Rates ad Asymptotic Equipartitio Sectio 29. itroduces the etropy rate the asymptotic etropy per time-step of a stochastic process ad shows that it is well-defied; ad similarly for iformatio,
More informationUniversity of Colorado Denver Dept. Math. & Stat. Sciences Applied Analysis Preliminary Exam 13 January 2012, 10:00 am 2:00 pm. Good luck!
Uiversity of Colorado Dever Dept. Math. & Stat. Scieces Applied Aalysis Prelimiary Exam 13 Jauary 01, 10:00 am :00 pm Name: The proctor will let you read the followig coditios before the exam begis, ad
More informationSolution to Chapter 2 Analytical Exercises
Nov. 25, 23, Revised Dec. 27, 23 Hayashi Ecoometrics Solutio to Chapter 2 Aalytical Exercises. For ay ε >, So, plim z =. O the other had, which meas that lim E(z =. 2. As show i the hit, Prob( z > ε =
More informationQuantile regression with multilayer perceptrons.
Quatile regressio with multilayer perceptros. S.-F. Dimby ad J. Rykiewicz Uiversite Paris 1 - SAMM 90 Rue de Tolbiac, 75013 Paris - Frace Abstract. We cosider oliear quatile regressio ivolvig multilayer
More informationMath 113 Exam 3 Practice
Math Exam Practice Exam will cover.-.9. This sheet has three sectios. The first sectio will remid you about techiques ad formulas that you should kow. The secod gives a umber of practice questios for you
More informationECE 901 Lecture 12: Complexity Regularization and the Squared Loss
ECE 90 Lecture : Complexity Regularizatio ad the Squared Loss R. Nowak 5/7/009 I the previous lectures we made use of the Cheroff/Hoeffdig bouds for our aalysis of classifier errors. Hoeffdig s iequality
More information