Hi AllI am trying to apply right now GARCH model to my time series of trade prices, but this last is not equally spaced in time since it comes from intraday trades.What to do?Can I use it anyway substituting the time series with one having more samples like the sample below?EXAMPLE:Original Time seriesTs(0), Ts(4), Ts(5)..New Time Seriests'(0) ts'(1) ts'(2) ts'(3) ts'(4) ts'(5)..having ts'(1) , ts'(2) and ts'(3) set to the value of Ts(0)?Is this approach correct for GARCH modeling?ThanksMnstn

P.S. I am really newbie. My question was more focused NOT on the modeling of the intraday trade prices (where other things like microstructure come to play), but more on the fact that the time series is not equally spaced in time. In this case what can I do? I read something that sounds like using a time series at an higher resolution ( like I wrote above ), modeling my Garch and downsample the results. Could it be good or should I go/investigate for some other approach?

No financial time series is equaly spaced in time. If you think about it you realise that you have various holidays, weekends and so on that mean you have no publicly traded prices. This also means you may have a price from one exchange but not from all, for certain dates. Then you'll also have overlap of trading time, with say London and NY opening and closing at different times. This assuming you still do the academic thing of only using some of the information. Even then you can use, opening, closing, highest, lowest, and even turnover. A lot of the work is in cleaning up the data to something you actually can use, that said I know a lot of teachers just bail on this, they make assignment with pre-packaged nicely behaved price-data to avoid spending time on Q on how to find, download, and treat data. Its still a bit of an art not a science (or you could have a package do it automatically).