IMeanVarianceOptimizer<T>

Interfaces · AiDotNet.Finance.Interfaces

A closed-form mean-variance portfolio optimizer: the analytic (training-free) weight solutions.

Known implementations: MarkowitzOptimizer<T>

For Beginners

“Mean-variance” optimization divides money across assets by trading off expected return against risk. These are the textbook closed-form answers; the default implementation is the standard Markowitz solver.

How It Works

This is a customization point, not a trainable model — unlike IPortfolioOptimizer (the neural, data-fit family), this exposes the classic closed-form Markowitz solutions a portfolio model can use directly as a baseline or warm start. Consumers default to MarkowitzOptimizer but can substitute their own analytic solver (shrinkage covariance, constrained QP, …).

Methods

Method Summary
Vector<T> MinimumVariance(Matrix<T> covariance) Global minimum-variance fully-invested weights (expected returns ignored).
Vector<T> Tangency(Vector<T> expectedReturns, Matrix<T> covariance, T riskFreeRate) Tangency (maximum-Sharpe) fully-invested weights for the given risk-free rate.
Vector<T> TargetReturn(Vector<T> expectedReturns, Matrix<T> covariance, T targetReturn) Minimum-variance fully-invested weights achieving a target expected return.