CODEWITHSUNDEEP
rxts
nesvarbu
nesvarbu

Reputation: 1016

Starting minute in minute data aggregation in xts

I have a question about xts to.hourly and split methods. It seems that it assumes that if my timestamp is 12:00 that it means a minute 12:00-12:01. However my data provider has 11:59-12:00 in mind. I didn't find any parameters on this. Is the only solution to simply lag my time series by one minute?

Upvotes: 0

Views: 76

Answers (1)

Joshua Ulrich
Joshua Ulrich

Reputation: 176648

Your question is actually about the endpoints function, which is what to.hourly, split.xts, and several other functions use to define intervals.

The endpoints function "assumes" that your timestamps are actual datetimes, and a time of 12:00:00 is the beginning of the 12 o'clock hour. If your data have 1-minute resolution, a time of 12:00:00 falls in the interval of 12:00:00.0-12:00:59.999.

There is no parameter you can change to make xts behave as if a time of 12:00:00 means anything other than what it actually is.

If you're certain that your data provider is putting a timestamp of 12:00 on data that occur in the interval of 11:59:00.0-11:59:59.999, then you might be able to simply subtract a small value from the xts object index.

# assuming 'x' is your xts object
.index(x) <- .index(x) - 0.001

That said, you need to think carefully about doing this because making a mistake can cause you to have look-ahead bias.

Upvotes: 1

Related Questions