Logo Questions Linux Laravel Mysql Ubuntu Git Menu
 

R xts: generating 1 minute time series from second events

Tags:

r

xts

I have an xts sequence of stock trade events that I want to process to generate 1 minute OHLC time series. For instance this set of trades:

Timestamp   Price  Size
9:30:00.123 12.32  200
9:30.00.532 12.21  100
9:30.32.352 12.22  500
9:30.45.342 12.35  200

Should result in the 9:30:00 record:

Timestamp Open  High  Low   Close
9:30:00   12.32 12.35 12.21 12.35

The way I approached this is to split the original trade series by the minute:

myminseries = do.call(rbind, lapply(split(mytrades, "minutes"), myminprocessing))

This produce the records I want but there is a problem: if the stock doesn't have any trade in a given minute, I will miss that minute record entirely. What I want instead is to have an all 0s record for the missing trades minute. For example, if there isn't any trade at 9:31:00 I should have:

Timestamp  Open  High  Low   Close
9:30:00    12.32 12.35 12.21 12.35
9:31:00    0     0     0     0
9:32:00    12.40 12.42 12.38 12.42

How can I backfill the 1 minute series? Or should I use a completely different approach than split()?

like image 799
Robert Kubrick Avatar asked Jan 17 '23 04:01

Robert Kubrick


1 Answers

If there are no trades within a given minute, to.minutes "will miss that minute record entirely" also. You can get around that by merging with a zero-width, strictly regular xts series.

## Make sample data
> x <- xts(cumsum(rnorm(600, 0, 0.2)), Sys.time() - 600:1) # 10 minutes of secondly data
> # remove all data from a couple different minutes
> x['2012-03-19 17:33'] <- NA
> x['2012-03-19 17:35'] <- NA
> x <- na.omit(x)
> 
> ## Convert to minutes
> xm <- to.minutes(x)
> head(xm)
                      x.Open   x.High       x.Low    x.Close
2012-03-19 17:31:59 0.1945049 1.661000 -0.35943057  1.6610000
2012-03-19 17:32:59 1.7283877 1.728388 -0.69288918  1.1398868
2012-03-19 17:34:59 2.0529582 2.603881 -0.80532315 -0.8053232
2012-03-19 17:36:59 0.5314270 1.189609 -0.94996548  0.5807342
2012-03-19 17:37:59 0.3761700 1.943363  0.04046976  0.9101720
2012-03-19 17:38:59 1.0614807 1.722110 -0.22147145  1.4075637
> axm <- align.time(xm) #align times to begining of next period
> 
> # to make strictly regular, create an xts object that has values for each minute
> tmp <- xts(, seq.POSIXt(start(axm), end(axm), by='min'))
> out <- cbind(tmp, axm)
> out
                       x.Open      x.High       x.Low     x.Close
2012-03-19 17:32:00  0.19450494  1.66100005 -0.35943057  1.66100005
2012-03-19 17:33:00  1.72838773  1.72838773 -0.69288918  1.13988679
2012-03-19 17:34:00          NA          NA          NA          NA
2012-03-19 17:35:00  2.05295818  2.60388093 -0.80532315 -0.80532315
2012-03-19 17:36:00          NA          NA          NA          NA
2012-03-19 17:37:00  0.53142696  1.18960858 -0.94996548  0.58073422
2012-03-19 17:38:00  0.37616997  1.94336348  0.04046976  0.91017202
2012-03-19 17:39:00  1.06148070  1.72211018 -0.22147145  1.40756366
2012-03-19 17:40:00  1.28437005  1.28437005 -0.62691689 -0.62691689
2012-03-19 17:41:00 -0.56820166  0.90339983 -0.77554869  0.26101945
2012-03-19 17:42:00 -0.07443971 -0.07443971 -0.07443971 -0.07443971
> na.locf(out)
                       x.Open      x.High       x.Low     x.Close
2012-03-19 17:32:00  0.19450494  1.66100005 -0.35943057  1.66100005
2012-03-19 17:33:00  1.72838773  1.72838773 -0.69288918  1.13988679
2012-03-19 17:34:00  1.72838773  1.72838773 -0.69288918  1.13988679
2012-03-19 17:35:00  2.05295818  2.60388093 -0.80532315 -0.80532315
2012-03-19 17:36:00  2.05295818  2.60388093 -0.80532315 -0.80532315
2012-03-19 17:37:00  0.53142696  1.18960858 -0.94996548  0.58073422
2012-03-19 17:38:00  0.37616997  1.94336348  0.04046976  0.91017202
2012-03-19 17:39:00  1.06148070  1.72211018 -0.22147145  1.40756366
2012-03-19 17:40:00  1.28437005  1.28437005 -0.62691689 -0.62691689
2012-03-19 17:41:00 -0.56820166  0.90339983 -0.77554869  0.26101945
2012-03-19 17:42:00 -0.07443971 -0.07443971 -0.07443971 -0.07443971

Or, if you really want zeros when there are no values, you could do out[is.na(out)] <- 0

like image 181
GSee Avatar answered Jan 31 '23 08:01

GSee