Simple R task: divide specified columns by 1000 at specified rows

Question

I have an OHLC array of stock quotes that I want to process.

             Open      High      Low        Close         Volume
2003-01-05   6111.01   6145.00   6102.70    6145.00         956
2003-01-08   6145.00   6190.00   5960.00    6135.05        8771
2003-01-09   6120.01   6250.00   6120.00    6225.00       10579
2003-01-10   6240.00   6285.00   6225.10    6261.00        8882
2003-01-13   6231.00   6325.00   6231.00    6270.00        8015
2003-01-14   6279.00   6295.00   6180.01    6190.00        8461

The company made a split @ given date, so I need to divide all open,high,low,close columns before that date by 1000. As I am learning R basics now I want to figure out nice R solution for this task. The best piece of code I've managed to code is (cant find out how to apply to given cols, stock$Open doesn't work):

apply(stock, 2, function(stock) stock/((index(stock)<"2007-07-20")*1000) )

However, the results are strange, many of them are inf:

2006-10-26       Inf       Inf       Inf        Inf         Inf
2006-10-27       Inf       Inf       Inf        Inf         Inf
2006-10-30       Inf       Inf       Inf        Inf         Inf
2006-10-31       Inf       Inf       Inf        Inf         Inf
2006-11-01       Inf       Inf       Inf        Inf         Inf
2006-11-02       Inf       Inf       Inf        Inf         Inf
2006-11-03       Inf       Inf       Inf        Inf         Inf
2006-11-07       Inf       Inf       Inf        Inf         Inf

Many thanks in advance!

Answer 1

I'm not familiar with OHLC arrays, but assuming that index method works:

relevantRows<-index(stock) < "2007-07-20"

Once you've got a vector holding all the relevant rows (in fact a logical vector that holds TRUE for the rows that should be changed), you can probably use this simply like this:

stock$Open[relevantRows]<-stock$Open[relevantRows]/1000

It is possible (depending on the internals of OHLC arrays), that even this works:

stock[relevantRows, c("Open", "High", "Low", "Close")]<-stock[relevantRows, c("Open", "High", "Low", "Close")]/1000

Answer 2

If the date is not before 20/7/2007, then (index(stock)<"2007-07-20") is FALSE and so (index(stock)<"2007-07-20")*1000 comes out as zero. Your Inf values are a result of dividing by zero.

You could try this:

stock[index(stock) < "2007-07-20", -5] <- stock[index(stock) < "2007-07-20", -5] / 1000

Answer 3

You can use the adjRatios function in the TTR package to do this. It looks like you already have an xts object, so that's what I use:

library(quantmod)
x <- structure(c(6111.01, 6145, 6120.01, 6240, 6231, 6279, 6145, 6190, 
6250, 6285, 6325, 6295, 6102.7, 5960, 6120, 6225.1, 6231, 6180.01, 
6145, 6135.05, 6225, 6261, 6270, 6190, 956, 8771, 10579, 8882, 
8015, 8461), .Dim = c(6L, 5L), .Dimnames = list(NULL, c("Open", 
"High", "Low", "Close", "Volume")), index = structure(c(1041746400, 
1042005600, 1042092000, 1042178400, 1042437600, 1042524000), tzone = "",
tclass = "Date"), class = c("xts", "zoo"), .indexCLASS = "Date", .indexTZ = "")
s <- xts(1/1000,as.Date("2003-01-10"))
r <- adjRatios(s,,Cl(x))
OHLC(x) * drop(r[,"Split"]) * drop(r[,"Div"])

If you're using data from Yahoo Finance, then you can use the adjustOHLC function in quantmod to automatically pull split and dividend data from Yahoo and adjust the series. See ?adjustOHLC for more options.