Portfolio optimization typically requires an estimate of a covariance matrix of asset returns. There are many approaches for constructing such a covariance matrix, some using the sample covariance matrix as a starting point. This package provides implementations for two such methods: random matrix theory and shrinkage estimation. Each method attempts to clean or remove noise related to the sampling process from the sample covariance matrix.

Documentation

Manual: tawny.pdf
Vignette: None available.

Maintainer: Brian Lee Yung Rowe <r at zatonovo.com>

Author(s): Brian Lee Yung Rowe

Install package and any missing dependencies by running this line in your R console:

install.packages("tawny")

Depends R (>= 3.0.0)
Imports lambda.r(>=1.1.6), lambda.tools, futile.logger(>=1.3.7), futile.matrix(>=1.2.1), tawny.types(>=1.1.2), zoo, xts, PerformanceAnalytics, quantmod
Suggests RUnit
Enhances
Linking to
Reverse
depends
Reverse
imports
Reverse
suggests
Reverse
enhances
Reverse
linking to

Package tawny
Materials
URL
Task Views Finance
Version 2.1.6
Published 2016-07-10
License GPL-3
BugReports
SystemRequirements
NeedsCompilation no
Citation
CRAN checks tawny check results
Package source tawny_2.1.6.tar.gz