PortfolioBacktest: portfolio building and backtesting
In raphael210/RFactorModel: A quantative tool for investment by factor Model

Description Usage Arguments Details Value Note Author(s) See Also Examples

portfolio building and backtesting

getPort(TSF, topN = NA, topQ = NA, sectorNe_pick = NULL, force_in = 0,
  buffer_keep = 0, buffer_rate = 0, init_port = NULL, dir = c("long",
  "short"))

addwgt2port(port, wgtType = c("eq", "fv", "fvsqrt", "custom", "ffv",
  "ffvsqrt"), sectorNe_wgt = NULL, wgtbmk = "EI000300", max_wgt = NULL,
  tolerance = 0.2)

port_limitwgt(port, max_wgt = NULL)

getPort_throughout(TSF, topN = NA, topQ = NA, sectorNe_pick = NULL,
  force_in = 0, buffer_keep = 0, buffer_rate = 0, init_port = NULL,
  wgtType = "eq", sectorNe_wgt = NULL, wgtbmk = "EI000300",
  max_wgt = NULL, dir = c("long", "short"))

port.backtest(port, holdingEndT = Sys.Date(), fee.buy = 0, fee.sell = 0,
  dir = c("long", "short"), rtn_get = c("loop", "whole", "simple"),
  silence = FALSE)

`TSF`	a TSF object or a TSFR object
`topN`	an integer vector with 2 or 1 elements, giving the rank range of the assets to be selected into the portfolio. If containing only 1 element, the the top rank of 1 will be added automatically.
`topQ`	a numeric vector with 2 or 1 elements, giving the percentage range of the assets to be selected into the portfolio. If containing only 1 element, the top percentage of 0 will be added automatically.
`sectorNe_pick`	NULL, "existing", or a sectorAttr.
`force_in`	a numeric between 0 and 1. eg. 0.1 means that stock with rank less than topN10% or pct less than topQ10% will be incorporated coercively.
`buffer_keep`	a numeric greater than 0. eg. 0.1 means that stock with rank less than topN110% (`topN(1+buffer_keep)`) or pct less than topQ*110% will be kept coercively.
`buffer_rate`	a numeric between 0 and 1
`init_port`	a charactor vector of stockIDs.
`dir`	a character string "long" or "short". In `getPort`, if "short",select from top to bottom, and vice versa.
`wgtType`	a character string, giving the weighting type of portfolio,which could be "eq"(equal),"fv"(floatValue),"fvsqrt"(sqrt of floatValue) or "custom".
`sectorNe_wgt`	NULL, "existing", or sectorAttr.
`wgtbmk`	a index ID.
`max_wgt`	a numeric or NA, giving the maximum of weight which could be set on a single stock. If NA(the default value), with no limit, return `port` itself without doing anything.
`tolerance`	a numeric, only used when `sectorNe_wgt` is not null, Giving the tolerance of absent sectors weights in the portfolio.
`holdingEndT`	a Date object, giving the ending date of the holding portfolio
`fee.buy`	giving the buying fee of each assets. See `Return.backtesting` for detail
`fee.sell`
`rtn_get`	a character string of "loop","whole","simple". "loop" and "whole" get the dailyrtn, where as "simple" get the periodrtn.
`dir`	a character string of "long" or "short". In `port.backtest`, if "long", the port will be look as asset, the fee will be cut from the asset; if "short", the port will be look as liability, the fee will be added to the liability.

getPort get the Port('portfolio') objects, subsetting from the TSF object.

addwgt2port add the weights to the Port object.

getPort_throughout, which is a wrapped function of getPort, addwgt2Port, port_limitwgt, get Port object from TSF, with further treatment.

port.backtest backtest the Port object. Get a PB("PortfolioBacktest") object.

getPort return a Port('portfolio') objects,which are of dataframe class containing at least 2 cols("date","stockID")

addwgt2port return a Port object which are of dataframe class containing at least 3 cols("date","stockID","wgt").

getPort_throughout return a Port object, which contain the col of 'wgt'.

port.backtest return a PB("PortfolioBacktest") object, a xts series of portfolio return, with the attr of 'turnover'(a xts series) and 'fee'(a vector).

Note that topN and topQ should at least have one and only have one.

Note that if use bufferring, the length of topN (or topQ) should be 1.

Ruifei.Yin

Return.backtesting

modelPar <- modelPar.default()
TSFR <- Model.TSFR(modelPar)
# -- good ports
Lport1 <- getPort(TSFR,20)
Lport2 <- getPort(TSFR,c(1,20)) # the same as Lport1
Lport3 <- getPort(TSFR,,0.2)
Lport4 <- getPort(TSFR,,c(0.1,0.2))
# -- bad ports
Sport1 <- getPort(TSFR,20,dir="short")
Sport2 <- getPort(TSFR,c(10,20),dir="short")
Sport3 <- getPort(TSFR,,0.2,dir="short")
Sport4 <- getPort(TSFR,,c(0.1,0.2),dir="short")
# -- N groups of ports
N <- 5
tmp <- seq(0,1,by=1/N)
groups <- cbind(tmp[-(N+1)],tmp[-1])
port.list <- apply(groups,1,function(x)getPort(TSFR,,x))
# -- with buffer_in_keep
ts <- getTS(as.Date(c("2016-03-31","2016-04-29")),indexID = "EI000016")
tsf <- getTSF(ts,"gf_lcfs",list("F000008"),factorDir = -1) 
pt <- getPort(tsf,20,force_in = 0.5,buffer_keep = 0.5)
pt2 <- getPort(tsf,20,force_in = 0.5,buffer_keep = 0.5,sectorNe_pick = defaultSectorAttr())
# -- with buffer_rate
pt3 <- getPort(tsf,20,buffer_rate =  0.5,sectorNe_pick =NULL)
pt4 <- getPort(tsf,topQ = 0.4,buffer_rate =  0.5,sectorNe_pick =NULL)
pt5 <- getPort(tsf,20,buffer_rate =  0.5,sectorNe_pick =defaultSectorAttr())
pt6 <- getPort(tsf,topQ = 0.4,buffer_rate =  0.5,sectorNe_pick =defaultSectorAttr())
# -- add wgt to port
port <- getPort(TSF,20)
port <- addwgt2port(port,"fv")
port <- addwgt2port(port,"fv",max_wgt=0.08)
# -- reduce the risk of concentration
port <- port_limitwgt(port,max_wgt=0.08)
# -- get Port object from TSF throughout
Port_throut <- getPort_throughout(TSF, topN=20, max_wgt=0.8, dir="long")
# -- backtest the Port object
PB <- port.backtest(port)
turnover <- attr(PB,"turnover")
fee <- attr(PB,"fee")