NeuralVaR<T>

Models & Types · AiDotNet.Finance.Risk

Neural Value-at-Risk (VaR) model for non-linear market risk assessment.

Inheritance: NeuralNetworkBase<T> → FinancialModelBase<T> → RiskModelBase<T> → NeuralVaR<T>

For Beginners

Value-at-Risk (VaR) is a way to answer the question: “What is the most I could lose on this investment tomorrow with 95% confidence?” Traditional methods often assume simple patterns, but this AI model “learns” from historical market crashes and complex trends to give a more realistic estimate of risk.

How It Works

NeuralVaR uses deep neural networks to estimate the potential loss of a portfolio under various market conditions, accounting for complex non-linear dependencies.

Reference: Barrera et al., “Statistical Learning of Value-at-Risk and Expected Shortfall”, 2026.

Example

// Define architecture for Value-at-Risk estimation (10 market risk factors)
var architecture = new NeuralNetworkArchitecture<double>(
    inputType: InputType.OneDimensional,
    taskType: NeuralNetworkTaskType.Regression,
    inputSize: 10, outputSize: 1);
            
// Training mode: neural network learns non-linear VaR from historical data
var model = new NeuralVaR<double>(architecture);
            
// ONNX inference mode: load pre-trained VaR model
var onnxModel = new NeuralVaR<double>(architecture, "neural_var.onnx");

Constructors

Constructor Summary
NeuralVaR(NeuralNetworkArchitecture<T> architecture, NeuralVaROptions<T> options, IGradientBasedOptimizer<T, Tensor<T>, Tensor<T>> optimizer, ILossFunction<T> lossFunction) Creates a NeuralVaR model in native mode for training.
NeuralVaR(NeuralNetworkArchitecture<T> architecture, string onnxModelPath, NeuralVaROptions<T> options, IGradientBasedOptimizer<T, Tensor<T>, Tensor<T>> optimizer, ILossFunction<T> lossFunction) Creates a NeuralVaR model using a pretrained ONNX model.

Methods

Method Summary
ModelMetadata<T> GetModelMetadata() Executes GetModelMetadata for the NeuralVaR.
T CalculateRisk(Tensor<T> input)
Tensor<T> AdjustForRisk(Tensor<T> action, T riskBudget)