Time series modelling of high frequency stock transaction data

摘要: This thesis comprises four papers concerning modelling of financial count data. Paper [1], [2] and [3] advance the integer-valued moving average model (INMA), a special case autoregressive (INARMA) class, apply models to number stock transactions in intra-day [4] focuses on long memory property time series data applying setting. [1] advances INMA stocks intraday The conditional mean variance properties are discussed extensions include, e.g., explanatory variables offered. Least squares generalized method moment estimators presented. In small Monte Carlo study feasible least estimator comes out as best choice. Empirically we find support for use long-lag Swedish series. There is evidence asymmetric effects news about prices transactions. introduces bivariate (BINMA) applies BINMA allows both positive negative correlations between shows that correlation always smaller than one an absolute sense. mean, covariance given. Model include suggested. Using AstraZeneca Ericsson B it found there Empirically, two vector (VINMA) model. VINMA counts. unconditional first second order moments obtained. CLS FGLS discussed. capable capturing within transaction frequency due macroeconomic related specific stock. spillover effect from larger B. develops account framework high its empirical application B, have fractional integration property.