I need to calculate differences of a time series (e.g. prices) for various periodicities (daily/weekly/monthly). In quantmod (and e.g. the tq_transmute wrapper in tidyquant) I can do something similar for arithmetic and logarithmic returns (using the function "periodReturn"). The nice effect here is that I get daily/monthly/weekly rows automatically. Is there a function that does something similar for pure differences?
I tried to search the documentation and did not find a suitable function.
Getting differences can be done by using the endpoints function from the xts library. Assuming you created an AAPL object using the getSymbols function:
getSymbols('AAPL',from='2019-01-01', to = '2019-05-31’)
Monthly differences:
monthlyDif <- AAPL$AAPL.Adjusted - lag(AAPL[endpoints(AAPL, on = 'months'),"AAPL.Adjusted"])
monthlyDif
AAPL.Adjusted
2019-01-31 NA
2019-02-28 7.392303
2019-03-29 16.735565
2019-04-30 10.678879
2019-05-30 -21.600189
Be aware that by using a combination of the on argument and the k argument you can also get differences from milliseconds to years or multiples like 2-week differences etc.
i.e. to get 2-week differences:
twoWeeksDif <- AAPL$AAPL.Adjusted - lag(AAPL[endpoints(AAPL, on = 'weeks',k = 2),"AAPL.Adjusted"])
twoWeeksDif
AAPL.Adjusted
2019-01-04 NA
2019-01-18 8.490769
2019-02-01 9.621521
2019-02-15 4.593429
2019-03-01 4.532547
2019-03-15 11.107224
2019-03-29 3.815307
2019-04-12 8.885773
2019-04-26 5.409180
2019-05-10 -6.336273
2019-05-24 -18.209992
2019-05-30 -0.669998