I am beginning to look at high frequency data sets. I have only ever dealt with regularly spaced data up until now so wondering what best practice is.
If I have two time series X and Y with Y being the dependent variable, if both time series are irregular is the only way to regress Y against X to perform some form of interpolation on both series?
Or is there a way to consider the time between data points as equal in duration and then perform a regression with X lagged against Y. I assume would have to consider that if you get a data point for X at time stamp 0 mics then the next data point in Y must occur after xx mics since it takes time for any response to occur in Y in relation to X.
Is there a good way to determine that duration of response time? or is this just an assumption?
