randport.eval {PortfolioProbe}R Documentation

Evaluate Each Random Portfolio

Description

Takes a random portfolio object and evaluates each individual portfolio. Any set of components of the output of trade.optimizer can be saved.

Usage

randport.eval(x, keep = c("results", "alpha.values", "var.values", 
	"utility.values"), subset = NULL, do.warn = FALSE, 
	additional.args = NULL, checkinput = TRUE, 
	envir = parent.frame(), FUN = NULL, ...)

Arguments

x

required. An object that is the result of a call to random.portfolio or random.portfolio.utility.

keep

character vector of component names of the output of trade.optimizer. If this is NULL, then the slightly ironic result is that all components are kept. Values are allowed to partially match the names of optimization object components.

subset

either NULL or a vector to subscript along the components of x.

do.warn

the control for the warnings coming from trade.optimizer. The default is FALSE, meaning all warnings are turned off that can be turned off. The two other likely values are: TRUE (turn on all warnings), and NULL (use the do.warn value that was in effect to start with). (A value of NULL here is not necessarily the same as NULL in trade.optimizer.)

This can be any value that the do.warn argument to trade.optimizer will accept.

additional.args

list of arguments to trade.optimizer either not in the random.portfolio call or that should be changed. In the latter case, argument names need to be given exactly the same as in the original – that is, with the same amount of abbreviation.

checkinput

a logical value. If TRUE, then a (fallible) check is made to see if there is a change in the value of price, variance, expected.return or existing between the current computation and the original computation. There is no warning about arguments that are in additional.args.

envir

the environment where the objects used to generate the random portfolios are going to be found. This is nonsensical to S+.

FUN

a function or a character string naming a function. Each portfolio trade object will be passed to the function.

...

additional arguments given to FUN.

Value

a list as long as x (or by what is indicated by subset if given).

If FUN is not given, each component is a list containing the components of the output of trade.optimizer listed in keep.

If FUN is given, then each component is the result of FUN applied to the portfolio trade object for the component.

Details

The objects that are involved (like prices and variances) need to be available in the current session, and presumably should be unchanged.

If you get a warning about a "checkinput difference" for one or more arguments, then you probably want to add those arguments to additional.args. Problems like this can arise because the arguments for the original computation were dependent on the iteration in a loop. Outside that original loop the value may be interpreted differently (or cause an error).

If neither expected returns nor variance are given (and a distance utility is not specified), then the utility is set to zero by maximizing expected returns that are all zero.

Bugs

There are limited circumstances when the utility.values component is confused (equal to zero) if expected.return can not be found and utility is not specified.

Revision

This help was last revised 2013 April 19.

See Also

random.portfolio, trade.optimizer, random.portfolio.utility.

Examples

randport1 <- random.portfolio(100, prices, variance=varian, 
	long.only=TRUE, bench.constr=c(spx=.04^2/252), 
	lin.constraints=cntrysect.conmat,
	lin.bounds=cntrysect.bounds, gross.value=1e6)

randeval1 <- randport.eval(randport1)

# look at linear constraints of the random portfolios
randlincon1 <- randport.eval(randport1, keep="con.realized")

# returns the above plus more
randsummary <- randport.eval(randport1, FUN=summary)

# vector of weight differences
rdist <- unlist(randport.eval(randport1, FUN = function(x)
    trade.distance(x, some.portf, prices=privec, scale="weight")))
[Package PortfolioProbe version 1.06 Index]