Skip to content

v1.1.0-rc.1

Pre-release
Pre-release

Choose a tag to compare

@GabrielMarquezMatte GabrielMarquezMatte released this 19 Nov 14:14

VolEngine 1.1.0-rc.1

Release candidate for 1.1.0. It adds a .NET binding and the EWMA (RiskMetrics) model. The C ABI only grew:
VE_ABI_VERSION stays 1, and code built against 1.0.0 keeps working.

Install

pip install --pre volengine==1.1.0rc1
dotnet add package VolEngine --version 1.1.0-rc.1

C++ wrapper:

FetchContent_Declare(volengine GIT_REPOSITORY https://github.com/GabrielMarquezMatte/VolEngine.git
                     GIT_TAG v1.1.0-rc.1 SOURCE_SUBDIR cpp)

Added

.NET binding: the VolEngine package on nuget.org (.NET 10).
It covers every feature of the C API:

  • univariate fits and grids over a panel
  • regressors and standard errors
  • information criteria and diagnostics
  • forecasts and simulation
  • DCC, aDCC, FDCC and Copula-DCC: fits, standard errors, forecasts and simulation

Specs and options are records. Results are immutable and own their native memory, so dispose them with using.
Errors are exceptions.

The package ships the native library for linux-x64, linux-arm64, linux-musl-x64, linux-musl-arm64, osx-arm64,
osx-x64 and win-x64, so no extra setup is needed. It works with trimming and Native AOT. The public API is stable
from 1.1.0 and follows semantic versioning. Guide: docs/guide/csharp.md.

using var context = new Context();
using var fit = context.Fit(new Spec { Model = Model.GjrGarch, Distribution = Distribution.Std }, returns);
Console.WriteLine($"BIC {fit.Bic():F4}; sigma in 10 steps {fit.Forecast(10).Sigma[^1]:F4}");

EWMA (RiskMetrics). It is VE_MODEL_EWMA in C, ve::Model::ewma in C++, Spec("ewma") in Python and
Model.Ewma in .NET.

  • The variance is σ²ₜ = λ σ²ₜ₋₁ + (1 − λ) ε²ₜ₋₁. This is iGARCH(1,1) with ω = 0.
  • lambda lies in [0, 1] and is estimated by maximum likelihood unless you fix it. fixed={"lambda": 0.94} gives
    classic RiskMetrics.
  • omega and alpha1 are reported as derived values: 0 and 1 − λ. Their standard errors are NaN, and the
    information criteria do not count them.
  • It accepts every mean, regressor, distribution, var_init, diagnostic, forecast (flat), simulation from a fit and
    DCC/copula margin that the other models accept.
  • Order (1, 1) only. Variance targeting and ve_simulate_spec are rejected, as for iGARCH.
  • Validated against arch's EWMAVariance.

Changed

  • An iGARCH fit with a free weight whose derived beta_q is within 1e-8 of 0 is now treated as on the boundary.
    Its standard error is NaN. Before, the Hessian step left the domain. The same rule applies to an EWMA lambda
    close to 1.

Assets

  • Native libraries for the seven platforms. On Windows there are libvolengine.dll, the .lib import library for
    MSVC and the .dll.a import library for MinGW.
  • volengine-abi-linux-x64.xml: the libabigail ABI dump.
  • volengine.spdx.json: the SBOM.

Feedback

This is a prerelease. Please report problems before 1.1.0 is final, especially with the NuGet package on your
platform.