On the Specification of Duration Between Price Changes and the Predictability of High Frequency Returns : An Application to the French CAC 40

摘要: This paper focuses on the predictability of duration between intraday price changes stocks from CAC 40 using a full year data, as well returns generated by these changes. It is argued that traders with different time horizons will look at series recorded intervals. Small variations be higher importance to short terme than longer term traders. We generate various reflect horizons. The specified model inspired Autoregressive Conditional Duration (ACD) Engle and Russell (1997). Estimated durations appear highly significant in predictiong future returns. These are economically important even after deduction bid-ask spread effect some strategies found profitable when tested 3 months' out-of-sample period. I also introduce threshold versions for return specifications, where estimates depend "state-of-the-world" defined thresholds obtained through cluster analysis. Simple clustering shown temove part dynamics captured ACD model.