Dynamic causality between intraday return and order imbalance in NASDAQ speculative top gainers

Abstract

This study explores dynamic conditional and unconditional causality relations between intraday return and order imbalance on extraordinary events. We examine intraday behaviour of NASDAQ speculative top gainers. In this study, we employ a regression model to examine intraday return-order imbalance behaviours. Moreover, we introduce a multiple-hypotheses testing method, namely a nested causality, to identify the dynamic relationship between intraday returns and order imbalances. We find order imbalance convey more information than trading volume does. While examining three intraday time regimes, we find the contemporaneous order imbalance-return effect is significant in the third sub-period, which implies that informed trading will take place in the afternoon. The size-stratified results show there is a negative relation between firm size and the order imbalance–return effect. The impact of the trading volume on the order imbalance–return effect is weaker than that of the firm size. Moreover, the volume-stratified results suggest that order imbalance be a better return predictor in small trading volume quartile and the order imbalance-based trading strategies are useful in the afternoon regime.

Notes

¹ The market makers would revise the price downward (upward) when there are excess sell (buy) orders.

² Lee et al. (2001) use six-minute intervals with each interval containing nearly 12 trades on average. Ekinci (2004) constructs five-minute intervals for an intra-day analysis of stocks with 27.3 trades per interval on average. For our sample period, there is only one day, we shorten the time interval. In addition, for NASDAQ dealers are required to report trades within 90 seconds, we use 90-second intervals to catch the intra-day seasonality.

³ The Island-ECN website is http://www.island.com. We would sign trades using Lee and Ready (1991) algorithm if we use the NYSE Trades and Automated Quotations (TAQ) databases. Unlike TAQ databases, the ‘Time and Sales’ database provided by Island-ECN has indicated the sign of trades.

⁴ When we use the number or the dollar volume of buyer-initiated trades minus that of seller-initiated trades to represent the order imbalance, the results are similar.

⁵ Trading is highest at the beginning and at the end of the day.

⁶ According to Ronald et al. (2005), transaction price is better than mid-point of bid–ask spread as a proxy of asset value.

⁷ The tests of conditional dynamic relations are based on the definitions defined in (17) and (18). The conditional tests incorporate past-order imbalances into the VAR system of order imbalances and returns, and therefore, account for the information impounded in past returns. Furthermore, the conditional tests allow us to better identify the dynamic relationship between returns and order imbalances in an inter-temporal setting.