Title: Predicting intraday return patterns based on overnight returns for the US stock market
Authors: Hao Li - University of Amsterdam (Netherlands) [presenting]
Cees Diks - University of Amsterdam (Netherlands)
Valentyn Panchenko - Univerisity of New South Wales (Australia)
Abstract: A new approach, cumulative regression (CumRe), is proposed in order to predict intraday financial return patterns conditional on observed overnight returns. Based on Trade and Quote data, we find evidence for dependence between overnight returns and subsequent intraday first and last half-hour return patterns for the S\&P 500 Exchange-Traded Fund for the time period from 2003 to 2013 with both statistical and economic significance. Our methodology allows studying the return patterns documented in the existing theoretical and empirical literature in more detail. Moreover, we find that both the first and the last half hours offer opportunities for day traders. Specifically, 20-minute after the market opens, and 30-minute before the closing are the best times for trading in terms of annualized returns, Sharpe ratios, and the difference between the Certainty Equivalent Returns.