GARCH Toolbox

Data Size and Quality

The size and quality of your return series affect the validity of your results. Because of this, The MathWorks recommends that you carefully examine your data prior to estimation. In particular, you should consider altering any missing or anomalous data points. For example, you can fill in missing data points, and remove or smooth anomalous ones. (See the section GARCH Limitations.)

In addition, GARCH volatility modeling typically requires at least a few hundred observations. Assuming daily data, one year's worth of data requires about 250 data points.

Boundary Constraints and Statistical Inferences

Function Reference