Table of Contents

Class VARMAModelOptions<T>

Namespace
AiDotNet.Models.Options
Assembly
AiDotNet.dll

Configuration options for Vector Autoregressive Moving Average (VARMA) models, which extend VAR models by incorporating moving average terms.

public class VARMAModelOptions<T> : VARModelOptions<T>

Type Parameters

T
Inheritance
VARMAModelOptions<T>
Inherited Members

Remarks

Vector Autoregressive Moving Average (VARMA) models extend Vector Autoregressive (VAR) models by incorporating moving average (MA) terms. While VAR models express each variable as a linear function of past values of itself and past values of other variables, VARMA models also include past error terms. This additional flexibility can lead to more parsimonious models and better forecasting performance, especially when the true data generating process includes moving average components. VARMA models are particularly useful for modeling and forecasting multiple interrelated time series, capturing both the autoregressive and moving average dynamics in the system. This class inherits from VARModelOptions and adds parameters specific to the moving average component of VARMA models.

For Beginners: VARMA models extend VAR models by including both past values and past errors.

When modeling multiple related time series:

  • VAR models use past values of all variables to predict future values
  • VARMA models add another component: past prediction errors

This additional component:

  • Captures patterns in the errors or "shocks" to the system
  • Can lead to more accurate and efficient models
  • Is particularly useful when the effects of shocks persist for multiple periods

Think of it like this:

  • VAR: "Tomorrow's values depend on today's and yesterday's values"
  • VARMA: "Tomorrow's values depend on today's and yesterday's values, plus how wrong our recent predictions were"

This class lets you configure the moving average component of VARMA models, while inheriting all the configuration options for VAR models.

Properties

MaLag

Gets or sets the lag order for the Moving Average (MA) component.

public int MaLag { get; set; }

Property Value

int

A non-negative integer, defaulting to 1.

Remarks

This property specifies the lag order for the Moving Average (MA) component in the VARMA model. The MA lag order, often denoted as q, determines how many past error terms (or "shocks") are included in the model for each variable. An MA lag of q means that the current values depend on the error terms from the current time point up to q time points in the past. A higher order allows the model to capture more complex patterns in the error terms but increases the number of parameters to estimate and the risk of overfitting. The default value of 1 provides a simple MA component suitable for many applications, capturing first-order effects of past errors. The optimal value depends on the autocorrelation structure of the residuals from a pure VAR model and the desired trade-off between model complexity and fit.

For Beginners: This setting controls how many past prediction errors influence the current values.

The Moving Average (MA) lag:

  • Determines how many previous prediction errors affect the current values
  • Helps the model capture the lingering effects of unexpected events or "shocks"
  • Higher values can model more complex patterns in how shocks persist

The default value of 1 means:

  • The model considers the immediate previous prediction error
  • This is sufficient for many time series with simple error structures

Think of it like this:

  • Higher values (e.g., 2 or 3): Can capture more complex patterns in how shocks persist
  • Lower values (e.g., 0): Effectively turns the model into a standard VAR model

When to adjust this value:

  • Increase it when the autocorrelation of residuals from a VAR model shows significant patterns
  • Keep at 1 for a simple VARMA model
  • Set to 0 if you want a pure VAR model
  • Consider using information criteria (AIC, BIC) to select the optimal order

For example, in financial time series where shocks often have lingering effects, you might increase this to 2 to better capture how market shocks persist over time.

Edit this page