conditional.sort: Conditional Portfolio Sort

In portsort: Factor-Based Portfolio Sorts

Description

Calculates out-of-sample mean sub-portfolio returns and the composition of each sub-portfolio using the conditional portfolio sorting method.

Usage

conditional.sort(Fa,Fb=NULL,Fc=NULL,R.Forward,dimA,dimB=NULL,dimC=NULL,type = 7)

Arguments

Fa	xts-object containing data for the first dimension of sort
Fb	xts-object containing data for the second dimension of sort (optional)
Fc	xts-object containing data for the third dimension of sort (optional)
R.Forward	xts-object containing forward returns
dimA	vector of break points between 0 and 1
dimB	vector of break points between 0 and 1 (optional)
dimC	vector of break points between 0 and 1 (optional)
type	pass-through parameter to the quantile function

Details

The conditional sort function sorts assets based on each factor (Fa to Fc) from low to high in a dependent fashion at each time t. Based on the sorted assets in each sub-portfolio at time t, mean out-of-sample sub-portfolio returns are computed for time t+1. After each dimension of sort, the subsequent sort is done only within each prior sorted sub-portfolio. Hence, the first factor that is sorted on yields greater influence on the overall sorting procedure. The function outputs out-of-sample returns for each sub-portfolio in columns and a list of the sub-portfolio constituents at each rebalancing point.

Value

returns	Out-of-sample sub-portfolio returns
portfolio	List of the sub-portfolio constituents over time

Note

The function implicitly handles NA/NaN or Inf values at each rebalancing point (at time t) by excluding them from the quantile function. Furthermore, if there are any NA, NaN or Inf values in the R.Forward object when computing out-of-sample returns, these are also excluded. The function outputs returns in columns. For example, if a double sort is conducted with both Fa and Fb including 3 breakpoints (a 3v3) sort, column 1 will contain out-of-sample returns for the 'Low-Low' sub-portfolio, column 4 will contain out-of-sample returns for the 'Mid-Low' sub-portfolio whilst column 9 will contain the 'High-High' sub-portfolio returns.

Author(s)

Jonathan Spohnholtz and Alexander Dickerson

Examples

# Load the included data
library(portsort)
data(Factors)

# Specifiy the sort dimension - in this case, a double sort on lagged returns and Bitcoin volumes
# with 4 breakpoints (a 4v4 sort)
dimA = c(0,0.25,0.5,0.75,1)
dimB = c(0,0.25,0.5,0.75,1)

# Specify the factors for the double sort
# Lagged returns, lagged volumes are stored in the Factors list

R.Forward = Factors[[1]]; R.Lag = Factors[[2]]; V.Lag = Factors[[3]]

# Subset the data from late 2017
R.Forward = R.Forward["2017-12-01/"]
R.Lag = R.Lag["2017-11-30/2018-09-05"]
V.Lag = V.Lag["2017-11-30/2018-09-05"]

Fa = R.Lag
Fb = V.Lag

# Conduct a conditional sort
sort.output <- conditional.sort(Fa,Fb,Fc=NULL,R.Forward = R.Forward,dimA = dimA,dimB = dimB)

portsort documentation built on May 2, 2019, 6:36 a.m.