[R-SIG-Finance] Extracting OHLC from trade price series

[Gabor Grothendieck]

On Wed, Feb 20, 2008 at 11:46 AM, Jeff Ryan <jeff.a.ryan at gmail.com> wrote:
> I wanted to clarify my comment with respect to timeDate:
>
> >
> >  The GMT issue has been discussed here many times over - try
> >  Sys.setenv(TZ='GMT') for starters.  You are probably better off _not_
> >  using timeDate - but I wrote xts so I wouldn't have to - to each his
> >  own.
> >
> >
>
> The 'timeDate' class itself is quite worthwhile.  There are many
> methods that seem to be available that can make life easier.  My point
> was more to the task at hand.  Simply subsetting an existing series,
> or aggregating to a lower periodicity doesn't require much beyond
> knowing time-stamps.
>
> If time-stamps are needed - POSIXct, Date, and chron are all up to the
> task.  It is entirely possible that 'timeDate' is equally useful, but
> my thinking is always to start simple and them progress to the more
> complicated only if necessary.
>
> For what it's worth, 'xts' can also handle indexing by 'timeDate'.
> See ?indexClass
>
> I'd love to hear a bit from the Rmetrics developers regarding the
> merits of 'timeDate'.  A comparison table of the different time/date
> classes would be very beneficial to many I suspect.  I'd be more than
> willing to try and put one together.

Date, chron and POSIXct are discussed and compared in the R News 4/1
help desk article.  There is a comparison table at the end of the article.