This paper examines high frequency stock returns with buy/sell signals. It demonstrates how such trading information could be utilized in a qualitative threshold framework to explain and predict the asymmetric behaviour of intraday stock returns. The study discovers that the buyer-dominating regime is consistently associated with negative returns, while the seller-dominating regime is consistently associated with positive returns. This is consistent with a suggestion of using the sign of the net buy/sell trading volume as the threshold indicator. Furthermore, the model renders better predicting power than that produced by a pure generalized autoregressive conditional heteroscedasticity model. Most interestingly, these results are quite robust across all 12 actively traded stocks on the Australian Stock Exchange that have been examined, and hence provide strong support for the potential usefulness of buy/sell signals and the qualitative threshold model in analysing the dynamics of high frequency financial asset returns.