[R-SIG-Finance] Defining many future_series() at once in FinancialInstruments

[G See]

Markus,

Thanks for the report.

The part that is slow is when it is searching the instrument
environment to see if the future_series you are trying to define has
already been defined.  I managed to more than double the speed of that
search following a suggestion from Brian to use unlist and regular
expressions.

However, the real speed up is to skip that step.  So, I added an
"overwrite" argument.  Now, future_series will not check to see if
there is already a future_series defined unless you call it with
overwrite=FALSE.

Patched at Rev 913.

Timings of your test below.

## Before patch
> system.time(future_series(all_contracts[1:200]))
   user  system elapsed
  2.600   0.000   2.595
> system.time(future_series(all_contracts[201:800]))
   user  system elapsed
 21.330   0.000  21.351
> system.time(future_series(all_contracts[801:2000]))
   user  system elapsed
104.950   0.000 105.041

> ## Update 1: unlist and regexpr instead of for loop
> system.time(future_series(all_contracts[1:200]))
   user  system elapsed
  1.160   0.000   1.158

> system.time(future_series(all_contracts[201:800]))
   user  system elapsed
  8.810   0.000   8.816

> system.time(future_series(all_contracts[801:2000]))
   user  system elapsed
 44.030   0.000  44.071

> ## New version
> system.time(future_series(all_contracts[1:200]))
   user  system elapsed
  0.180   0.000   0.176

> system.time(future_series(all_contracts[201:800]))
   user  system elapsed
  0.520   0.000   0.522

> system.time(future_series(all_contracts[801:2000]))
   user  system elapsed
  1.050   0.000   1.054


Regards,
Garrett


On Tue, Jan 17, 2012 at 4:55 AM, varcovar <varcovar at live.com> wrote:
> Hi guys,
>
> I have 100 Futures symbols that are defined using the future() function from
> the FinancialInstrument package. I want to define their respective contracts
> using the future_series() function. I have about 15000 contracts to define.
> The problem here is that the time needed to define a contract with
> future_series() increases exponentially with the number of contracts defined
> (it is very fast for the first contract, and then goes slower and slower).
>
> Here is a reproducible example. I build a vector of 104 market symbols and
> then a vector of 19136 contracts (184 contracts for each market symbol).
> Then I calculate the time needed to define the first 200 contracts, and then
> 600 contracts. While it takes about 2.76 seconds on my desktop to define the
> first 200 contracts, it takes more than 23 seconds for the next 600
> contracts. At this increasing rate, it would take an eternity to define all
> those 19136 contracts.
>
> library(quantstrat)
> ## create a vector of 104 symbols
> symbols <- NULL
> for (i in 1:4)
>    symbols <- c(symbols, paste(LETTERS[(2 * i):26], LETTERS[1:26], sep =
> ""))
> currency("USD")
> ## create a vector containing all contracts for the 104 symbols (thus 19136
> contracts)
> all_contracts <- NULL
> for (i in 1:length(symbols)) {
>    ## define futures
>    future(primary_id = symbols[i], currency = "USD", multiplier = 100,
>        tick_size = 0.1, month_cycle = "F,G,H,J,M,Q,V,Z")
>    ## create contracts series
>    contracts <- build_series_symbols(
>        roots = data.frame(
>            primary_id  = getInstrument(symbols[i])$primary_id,
>            month_cycle = getInstrument(symbols[i])$month_cycle),
>        yearlist = 1990:2012)
>    ## format contracts symbols
>    contracts <- format_id(contracts, format = "CYY")
>    ## append new contracts to all contracts
>    all_contracts <- c(all_contracts, contracts)
> }
> ## calculate time needed to define contracts
> system.time(future_series(all_contracts[1:200]))
> system.time(future_series(all_contracts[201:800]))
>
> Once all the future_series are defined, I expect to save them using the
> saveInstruments() function for later use. Any suggestion for making this
> process faster? I really appreciate your help! Thank you.
>
> Markus Douglas
>
> --
> View this message in context: http://r.789695.n4.nabble.com/Defining-many-future-series-at-once-in-FinancialInstruments-tp4302922p4302922.html
> Sent from the Rmetrics mailing list archive at Nabble.com.
>
> _______________________________________________
> R-SIG-Finance at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-sig-finance
> -- Subscriber-posting only. If you want to post, subscribe first.
> -- Also note that this is not the r-help list where general R questions should go.