Identify Conditional Heteroscedacity
TimeSeriesModelFit automatically checks for conditional heteroscedacity in data and fits ARCH/GARCH models to data.
Create a time series of daily returns on Starbucks Corp. stock.
show complete Wolfram Language inputhide input
Out[3]= | ![](HTMLImages.en/identify-conditional-heteroscedacity/O_8.png) |
Compute the autocorrelation function.
show complete Wolfram Language inputhide input
Out[5]= | ![](HTMLImages.en/identify-conditional-heteroscedacity/O_9.png) |
Test for autocorrelation in the sequence of returns.
Out[6]= | ![](HTMLImages.en/identify-conditional-heteroscedacity/O_10.png) |
The returned time series is not autocorrelated, but its square is.
Out[8]= | ![](HTMLImages.en/identify-conditional-heteroscedacity/O_11.png) |
Out[9]= | ![](HTMLImages.en/identify-conditional-heteroscedacity/O_12.png) |
TimeSeriesModelFit determines the GARCH family as the best fit for the data.
Out[10]= | ![](HTMLImages.en/identify-conditional-heteroscedacity/O_13.png) |
Find the fitted process.
Out[11]= | ![](HTMLImages.en/identify-conditional-heteroscedacity/O_14.png) |
The model residuals appear uncorrelated.
Out[12]= | ![](HTMLImages.en/identify-conditional-heteroscedacity/O_15.png) |
Out[13]= | ![](HTMLImages.en/identify-conditional-heteroscedacity/O_16.png) |
Use TimeSeriesModel to compute confidence intervals of future forecast.
Out[15]= | ![](HTMLImages.en/identify-conditional-heteroscedacity/O_17.png) |