[R-SIG-Finance] Quantstrat - external parameters

[Brian G. Peterson]

On 08/25/2012 11:35 PM, Jaeyoung Ahn wrote:
> First of all, I appreciate all the hard works the developers have poured
> for this package. Is there anyone who could help me with this?
>
> I have searched this question and got to the following answer provided by
> Braverock.
> http://r.789695.n4.nabble.com/Parameters-setting-for-multiple-indicators-with-the-same-argument-names-td3793443.html
>
> I tried to follow his recommendation but couldn't get it work. My codes are
> as follows. It appears that only the first set of parameters are repeated
> over and over. External parameters (shortMA and longMA) don't seem like to
> be updated at all.

At this point I'm going to recommend that you look at

?setParameterDistribution
?setParameterConstraint
?applyParameters

and demo files luxor.3.R and luxor.3.Parameters.R

Like your example below, these demos are for a modified 2-MA cross 
strategy taken from the Tomasini book.  The numbers are chapter numbers, 
as in the book.

Lines 40-43 in luxor.3.Parameters.R set the distributions for each MA, 
and then define the relationship that nFast be less than ('lt') nSlow.

If you want to run these demos yourself, you'll need to choose an 
appropriate parallel backend for foreach.  On Windows, I'd recommend 
doParallel as the easiest place to start.

At the time of the prior thread, referenced above, there was no demo 
code for this type of strategy.  Right now, I think you're better off 
starting with applyParameters than writing your own loop.

Te model you outline below should work, it is likely failing do to a 
misplaced quote() somewhere, though the exact cause isn't obvious to me 
without more study.

One problem that I see immediately is that you are deleting the 
portfolio in each loop iteration, so you would only see the last results 
anyway.  It seems that it would make more sense to create a portfolio 
for each loop index, i, if shortMA<longMA so that you can collect all 
the results.

However, applyParameters does all this for you, so it seems that you'd 
be better off studying these demos instead.

Regards,

    - Brian

> input_data <- read.csv('mktdata.csv')
> EURUSD <- xts(input_data[,3:6], as.POSIXct(input_data[,'Date'], tz='PST',
> format='%Y-%m-%d', drop.time=F))
> names(EURUSD) <- c("OPEN","HIGH","LOW","CLOSE")
>
> shortMA = 5
> longMA = 10
>
> initDate <- as.Date('2000-01-02') - 1
>
> stock.str <- 'EURUSD'
> strat.name <- 'MAX'
> initEq=1000000
>
> dummy <- initPortf(name=strat.name,symbols=stock.str, initDate=initDate)
> dummy <- initAcct(name=strat.name,portfolios=strat.name,
>                    initDate=initDate, initEq=initEq)
> initOrders(portfolio=strat.name,initDate=initDate)
> strat <- strategy(strat.name)
>
> *#indicators:*
> strat <- add.indicator(strategy = strat, name = "SMA", arguments =
> list(x=quote(Cl(mktdata)), *n=quote(shortMA))*,label= "ma_sn" )
> strat <- add.indicator(strategy = strat, name = "SMA", arguments =
> list(x=quote(Cl(mktdata)), *n=quote(longMA))*,label= "ma_ln")
>
> # signals:
> strat <- add.signal(strategy = strat,name="sigCrossover",
>                      arguments = list(columns=c("ma_sn","ma_ln"),
> relationship="gte"),
>                      label="ma_sn.gt.ma_ln")
> strat <- add.signal(strategy = strat,name="sigCrossover",
>                      arguments =
> list(column=c("ma_sn","ma_ln"),relationship="lt"),
>                      label="ma_sn.lt.ma_ln")
>
> # rules:
> strat <- add.rule(strategy = strat,name='ruleSignal',
>                    arguments = list(sigcol="ma_sn.gt.ma_ln",sigval=TRUE,
> orderqty=1000000,
>                                     ordertype='market',
> orderside='long',pricemethod='market'),type='enter',path.dep=T)
> strat <- add.rule(strategy = strat,name='ruleSignal',
>                    arguments = list(sigcol="ma_sn.lt.ma_ln",sigval=TRUE,
> orderqty=-1000000,
>                                     ordertype='market',
> orderside='long',pricemethod='market'),type='exit',path.deop=T)
>
>
> library(PerformanceAnalytics)
>
> ShortM <- c(5,10,15) #,20)
> LongM <- c(20, 40) #, 50,60,100)
>
> parm.comb <- expand.grid(ShortM=ShortM,LongM=LongM)
> num.opts <- nrow(parm.comb)
> res.vec <- rep(NA,num.opts)
> res.vec2 <- res.vec
>
>
> for(i in 1:num.opts) {
>
>    # initialize portfolio and orders
>    try(rm(list=ls(pos=.blotter),pos=.blotter),silent=TRUE)
>    try(rm(list=ls(pos=.strategy),pos=.strategy),silent=TRUE)
>    dummy <- initPortf(name=strat.name,symbols=stock.str, initDate=initDate)
>    initOrders(portfolio=strat.name,initDate=initDate)
>
>    # apply strategy
>    ShortMA <- parm.comb[i,"ShortM"]
>    LongMA <- parm.comb[i,"LongM"]
>
>    if (ShortMA<LongMA) {
>      out<-try(applyStrategy(strategy=strat, portfolios=strat.name,
> verbose=F, parameters=list(shortMA=ShortMA, longMA= LongMA)))
>
>      # calculate performance matric
>      dummy <- updatePortf(Portfolio=strat.name,
> Dates=paste('::',as.Date('2012-08-17'),sep=""))
>
>      trading.pl <- getPortfolio(strat.name)$summary$Net.Trading.PL
>      rets <- trading.pl/initEq
>      omega <- as.numeric(Omega(rets))
>      res.vec[i] <- omega
>
>    }
> }
>
>