HierarchicalRiskParity<T>

Models & Types · AiDotNet.Finance.Portfolio

Neural Hierarchical Risk Parity (HRP) for portfolio optimization.

Inheritance: NeuralNetworkBase<T> → FinancialModelBase<T> → PortfolioOptimizerBase<T> → HierarchicalRiskParity<T>

For Beginners

HRP is a modern portfolio construction technique that uses machine learning to cluster similar assets together and then allocates risk equally across these clusters. Unlike traditional Mean-Variance Optimization, it doesn’t require inverting a covariance matrix, making it robust to noise and stable even when assets are highly correlated.

Example

// Create a Hierarchical Risk Parity model for robust portfolio construction
var architecture = new NeuralNetworkArchitecture<double>(
    inputType: InputType.OneDimensional,
    taskType: NeuralNetworkTaskType.Regression,
    inputHeight: 252, inputWidth: 10, inputDepth: 1, outputSize: 10);
var model = new HierarchicalRiskParity<double>(architecture);
            
// Or load a pre-trained ONNX model for HRP allocation
var onnxModel = new HierarchicalRiskParity<double>(architecture, "hrp.onnx");

Constructors

Constructor Summary
HierarchicalRiskParity() Initializes a new instance of the HierarchicalRiskParity model.
HierarchicalRiskParity(NeuralNetworkArchitecture<T> architecture, HierarchicalRiskParityOptions<T> options, IGradientBasedOptimizer<T, Tensor<T>, Tensor<T>> optimizer, ILossFunction<T> lossFunction)
HierarchicalRiskParity(NeuralNetworkArchitecture<T> architecture, string onnxModelPath, HierarchicalRiskParityOptions<T> options, IGradientBasedOptimizer<T, Tensor<T>, Tensor<T>> optimizer, ILossFunction<T> lossFunction) Initializes a new instance from ONNX.

Methods

Method Summary
Vector<T> OptimizePortfolio(Tensor<T> marketData) Optimizes portfolio weights using HRP logic.