We test a candidate high-frequency investment strategy, which utilizes the time series of price, volume and a novel interaction term to forecast intra-day returns over a continuous 101 day period from January 4 to May 28, 2010. The strategy uses minute-level data calculate regression coefficients from one-day for the purposes of trading the following day, thereby avoiding data snooping bias. Finally regressing the daily returns against the Fama-French Four Market Factors reveals significant alphas for more than half of the traded stocks.
FRM Project-Simon Fraser University
Copyright is held by the author(s).
Member of collection