jorda and marcellino-modeling high-frequency fx data dynamics

Modeling High-Frequency FX Data Dynamics ∗ ` Oscar Jord`a University of California, Davis ojorda@ucdavis.edu Massimiliano Marcellino Universit`aBocconi massimiliano.marcellino@uni-bocconi.it February, 2002 [...]... Irregular No Yes No Yes No Yes No Yes Random Irregular 6.4773 6.4035 6.1091 6.2055 Random Irregular 0.0000 0.0000 0.0000 0.0000 Random Irregular 0.5351 0.5531 0.4454 0.4462 Random Irregular Yes Yes Yes Yes Random Irregular No No No Yes* Comments: The title “operational time” refers to the original data; “Deterministic, Fixed Interval” refers to aggregating the original data every 100 periods; “Deterministic,... closing of major trading centers for the USD-DM FX is indicated 28 Figure 2 Weekly Pattern of Quote Arrival in the USD-DM FX Market: Raw Data 300 Quotes 200 100 0 Monday Tuesday Wednes Thursd Friday Saturd Sunday Raw quote-arrival for a typical week in the USD-DM FX Market 29 Figure 3 Nonparametric Estimates of the Autoregressive Parameters of the USDDM FX Spread Model as a Function of the Intensity... Rate in the USD-DM FX Market: Seasonal Intraday Pattern 200 Quote Counts London Opens 160 120 New York Opens Frankfurt Opens Asia Closes Europe Closes 80 New York Closes Tokyo and Sydney 40 Open 0 Lunch in Asia 6 12 18 24 1 Trading Day 30-minute intervals The graph displays non-parametric estimates of the seasonal pattern of a typical business day in intervals of 30-minutes The opening and closing of major... Interval” refers to aggregating the original data every 100 periods; “Deterministic, Irregular” refers to aggregating according to a seasonal cycle that repeats itself every 100 periods; and “Random, Irregular” refers to true random intervals of aggregation with a seasonal pattern The * indicates the test was barely significant at the conventional 95% confidence level 26 Table 2 Comparing the Poisson Regression... PHI2 PHI3 6 4 2 0 -.2 50 100 150 200 250 Intensity: Quotes/half-hour Value of the autoregressive parameters φ1, φ2, and φ3 of the model for the USD-DM FX spread variable as a nonparametric function of the intensity of quote arrival 30 Figure 4 Nonparametric Estimate of the Mean USD-DM FX Spread as a Function of the Intensity of Quote Arrival per 30-minute Interval 1.2 Spread 1.0 0.8 0.6 0.4 0.2 0.0... K reteP ,kralC ]3[ sserP ytisrevinU egdirbmaC :egdirbmaC ,03 on ,shpargonoM yteicoS cirtemonocE ,ataD tnuoC fo sisylanA noissergeR )8991( idevirT K nivarP dna C A ,noremaC ]2[ sserP ytisrevinU drofxO :drofxO ,gnilledoM cirtemonocE emiT suounitnoC )0991( R A ,mortsgreB ]1[ References slevel ytivitca fo ytisnetni eht ni segnahc htiw detaicossa tnetnoc noitamrofni ni noitairav eht sthgilhgih dna llew... dnalreztiwS ,hciruZ ,setaicossA dna neslO ,sgnideecorP ,ecnaniF ni ataD ycneuqerF hgiH ni ”,sledoM )HCRA-RASI( HCRA dna RA decapS ylralugerrI“ )5991( kesaloP W dna S J ,iaP ]72[ srehsil -buP llewkcalB :drofxO ,yroehT erutcurtsorciM tekraM ,)5991( ,neeruaM ,araH’O ]62[ 774-354 ,21 ,seidutS laicnaniF fo weiveR ”,stekraM egnahcxE ngieroF ni stsoC noitceleS esrevdA dna noitnevretnI tnemnrevoG“ )0002( nardnelamiN . Modeling High-Frequency FX Data Dynamics ∗ ` Oscar Jord`a University of California, Davis ojorda@ucdavis.edu Massimiliano Marcellino Universit`aBocconi massimiliano.marcellino@uni-bocconi.it February,. w2 h5" alt="" 1. Introduction