High frequency data entry: statistical findings at high frequency

Abstract

We introduce some of the most common types of high-frequency financial data: tick-by-tick data, trade andquote data, order bookdata, andmarket member data. We describe the types of variables that are usually available in the most popular high-frequency financial databases. We discuss the issues related to the handling of these data, including cleaning protocols, timing issues, and issues related to data size. We then briefly consider the issues related to the stylized facts detected in the empirical analysis of high- frequency data. Specifically, we consider (i) the irregular temporal spacing of the events at high frequency and its relevance for the econometric modeling of financial variables, (ii) the discreteness of the financial variables under investigation, (iii) the problems related to proper definition of financial variables, (iv) their daily periodicity, that is, typical intraday patterns, (v) their temporal correlations, for example, the bid–ask bounce and long memory properties, and (vi) problems related to the specificity of the market structure and rules.