ABSTRACT
This chapter reviews some major methods and models in high-frequency econometrics. A trade is an exchange of assets between two parties. In principle, any data and other information generated from the trading activities are transactions data and have been used by traders since ancient times. The Epps effect is mainly caused by two features of high-frequency data. The first is the stickiness of inactive trades under high sampling frequency. Another cause of the Epps effect is asynchronicity, leading to situations where an unusually large movement occurs in the synchronized price differences of an asset but the other assets still have zero or small synchronized price differences. The chapter considers autoregressive conditional duration models for the intraday time series of durations. It describes econometric models that use Trade and Quote data to predict future volatilities for applications to portfolio optimization, derivatives pricing and risk management.
