0

I am currently facing a problem with my GARCH model, specifically with the forecasted results. It appears that all the forecasted values are turning out to be identical, which leads me to suspect that these results are inaccurate. I would greatly appreciate any suggestions or guidance on how to resolve this issue.

In order to demonstrate the problem more clearly, I have prepared a simple example where I encountered the same issue—finding that the forecasted values are all the same.

Thank you in advance for your assistance, and I look forward to any insights or recommendations you can provide to address this matter.

EXAMPLE:

datos=[3.4, 5,7, 6.3, 4.5, 3.2, 3.8, 6.7]

Define the GARCH model


model = arch_model(datos, vol='Garch', p=1, q=1)


Fit the GARCH model


result = model.fit()


Print model summary


print(result.summary())


Fit the model


resultado = model.fit()


Print the model summary


print(resultado.summary())


Forecast future observations


forecast_horizon = 10  # Specify the number of periods to forecast
forecasts = resultado.forecast(start=0, horizon=forecast_horizon)


Extract the forecasted values


forecasted_values = forecasts.mean[-1:].values.flatten()


print("Forecasted Values:")
print(forecasted_values)


Forecasted Values:
[4.98545041 4.98545041 4.98545041 4.98545041 4.98545041 4.98545041
 4.98545041 4.98545041 4.98545041 4.98545041]
Milton Bertin-Jones
  • 1
  • 2
    Welcome to CV. Since you’re new here, you may want to take our tour, which has information for new users. Please check How do I ask a good question? and edit your title and main body of the text. The main body should include the question (not just in the title). More information is also welcome. – T.E.G. Jun 21 '23 at 10:56
  • 1
    Under GARCH, you could have constant or time-varying conditional mean. You must have specified (perhaps implicitly) that the conditional mean is constant. See this thread. – Richard Hardy Jun 21 '23 at 13:44
  • We have no way to know what model you're actually fitting, because you didn't specify a library or even a language in which the arch_model function you're using can be found. But, I would expect its default to have a constant mean, which is consistent with what you're showing here, and not an error. You should know that, generally, the point of GARCH itself is to model time-varying volatility (i.e. the spread around the mean value), not so much to try to capture complex dynamics in the mean (although, there are commonly used extensions, like ARMA-GARCH, that try to do both). – Chris Haug Jun 25 '23 at 16:02

0 Answers0