Base Distance
PortfolioOptimisers.SimpleDistance — Typestruct SimpleDistance <: AbstractDistanceAlgorithm endSimple distance algorithm for portfolio optimization.
\[\begin{align} d_{i,\,j} &= \sqrt{\dfrac{1 - \rho_{i,\,j}}{2}}\,, \end{align}\]
where $d$ is the distance, $\rho$ is the correlation coefficient, and each subscript denotes an asset.
Related
PortfolioOptimisers.SimpleAbsoluteDistance — Typestruct SimpleAbsoluteDistance <: AbstractDistanceAlgorithm endSimple absolute distance algorithm for portfolio optimization.
\[\begin{align} d_{i,\,j} &= \sqrt{1 - \lvert\rho_{i,\,j}\rvert}\,, \end{align}\]
where $d$ is the distance, $\rho$ is the correlation coefficient, and each subscript denotes an asset.
Related
PortfolioOptimisers.LogDistance — Typestruct LogDistance <: AbstractDistanceAlgorithm endLogarithmic distance algorithm for portfolio optimization.
\[\begin{align} d_{i,\,j} &= -\log{\lvert\rho_{i,\,j}\rvert}\,, \end{align}\]
where $d$ is the distance, $\rho$ is the correlation coefficient, and each subscript denotes an asset.
Related
PortfolioOptimisers.CorrelationDistance — Typestruct CorrelationDistance <: AbstractDistanceAlgorithm endCorrelation distance algorithm for portfolio optimization.
\[\begin{align} d_{i,\,j} &= \sqrt{1 - \rho_{i,\,j}}\,, \end{align}\]
where $d$ is the distance, $\rho$ is the correlation coefficient, and each subscript denotes an asset.
Related
PortfolioOptimisers.VariationInfoDistance — Typestruct VariationInfoDistance{T1, T2} <: AbstractDistanceAlgorithm
bins::T1
normalise::T2
endVariation of Information (VI) distance algorithm for portfolio optimization.
VariationInfoDistance specifies the use of the Variation of Information (VI) metric, an information-theoretic distance based on entropy and mutual information.
Fields
bins: Binning strategy or number of bins. If an integer, must be strictly positive.normalise: Whether to normalise the VI distance to the range [0, 1].
Constructor
VariationInfoDistance(; bins::Union{<:AbstractBins, <:Integer} = HacineGharbiRavier(),
normalise::Bool = true)Keyword arguments correspond to the fields above.
Validation
- If
binsis an integer,bins > 0.
Examples
julia> VariationInfoDistance()
VariationInfoDistance
bins | HacineGharbiRavier()
normalise | Bool: trueRelated
PortfolioOptimisers.CanonicalDistance — Typestruct CanonicalDistance <: AbstractDistanceAlgorithm endCanonical distance algorithm for portfolio optimization.
Defines the canonical distance metric for a given covariance estimator. The resulting distance metric is consistent with the properties of the covariance estimator (relevant when the covariance estimator is MutualInfoCovariance).
| Covariance Estimator | Distance Metric |
|---|---|
MutualInfoCovariance | VariationInfoDistance |
LTDCovariance | LogDistance |
DistanceCovariance | CorrelationDistance |
StatsBase.CovarianceEstimator | SimpleDistance |
The table also applies to PortfolioOptimisersCovariance where ce is one of the aforementioned estimators.
When used with a covariance matrix directly, uses SimpleDistance.
Related
PortfolioOptimisers.AbstractDistanceEstimator — Typeabstract type AbstractDistanceEstimator <: AbstractEstimator endAbstract supertype for all distance estimator types in PortfolioOptimisers.jl.
All concrete types implementing distance-based estimation algorithms should subtype AbstractDistanceEstimator. This enables a consistent interface for distance-based measures (such as correlation distance, absolute distance, or information-theoretic distances) throughout the package.
Related
PortfolioOptimisers.AbstractDistanceAlgorithm — Typeabstract type AbstractDistanceAlgorithm <: AbstractAlgorithm endAbstract supertype for all distance algorithm types in PortfolioOptimisers.jl.
All concrete types implementing specific distance-based algorithms (such as correlation distance, absolute distance, log distance, or information-theoretic distances) should subtype AbstractDistanceAlgorithm. This enables flexible extension and dispatch of distance routines for use in portfolio optimization and risk analysis.
Related