rollingCorrelation {PerformanceAnalytics}R Documentation

rolling training period covariance/correlation

Description

This is a wrapper for providing n-period trailing correlations for the data provided.

Usage

rollingCorrelation(Ra, Rj, width, trim = TRUE, na.rm = FALSE, ...)

Arguments

Ra a vector, matrix, data frame, timeSeries or zoo object of asset returns
Rj a vector, matrix, data frame, timeSeries or zoo object of asset returns
width the number of periods over which a function is to be calculated. Use the value zero (0) to roll the statistic from inception
trim TRUE/FALSE, whether to keep alignment caused by NA's
na.rm TRUE/FALSE Remove NA's from the returns?
... any other passthru parameters

Details

Example: head(rollingCorrelation(manager.ts@Data[,1],edhec.ts@Data,n=12))

           Convertible Arbitrage    CTA Global    Distressed
2003-11-28             0.2591101     0.2762218     0.7516556
2003-12-31             0.2162078     0.2477113     0.7452179
2004-01-30             0.3918575     0.3489062     0.7562063
2004-02-27             0.5331404     0.3905645     0.7088004
2004-03-31             0.5730389     0.3010877     0.5694478
2004-04-30             0.5146946     0.3762283     0.4374524

Value

A data.table of n-period trailing correlations for each column in y.

Note

Inspired by fMultivar rollFun written by Diethelm Wurtz.

Assumes that Ra and Rj are sequenced exactly the same and are regular.

Currently assumes that BOTH Ra and Rj are provided

@todo: Allow either a matrix or data frame for 'Ra' or give both 'Ra' and 'Rj' like cov

Author(s)

Peter Carl

References

See Also

cov

Examples

#head(rollingCorrelation(manager.ts@Data[,1],edhec.ts@Data,n=12))

[Package PerformanceAnalytics version 0.9.4 Index]