Title: The impact of periodicity on volatility-volume relations
Authors: Yi Luo - Lancaster University (United Kingdom) [presenting]
Zhen Wei - Xi An Jiaotong-Liverpool University (China)
Abstract: Opening, lunch, and closing of financial markets induce a periodic component in the volatility of high-frequency returns. However, the intraday volume and number of trades also display a prominent U curve that is still left to be investigated. We propose to use the Seasonal-Trend Decomposition Procedure Based on Loess to estimate the periodic component in volume and number of trades. We find that accounting for periodicity improves the explanatory power of both volume and number of trades on realized variance. Besides, the relationship between the average absolute return and volume (number of trades as well) can be better modeled using the mixture of two linear regression models during the trading day. With more analysis on the posterior probabilities of the mixing components, the average intraday volume and the number of trades display a higher effect on the absolute return in the morning relative to the rest of the day. The observed patterns indicate the need to decompose and analyze the periodicity not only in the realized variance but also in the volume and number of trades.