Forecasting rescaled
Posted: Sun Jun 18, 2017 11:09 am
I've been trying to forecast (static) the colcap (colombian stock exchange index). I modelled the volatility to be an egarch(1,1) for an ar(1)
The program runs, however the result is around 10 times smaller than it should. It should give me a value in a range of -0.25 up to 0.25, however my forecast oscillates around -0.025 and 0.025.
When i multiply the forecasted series times 8 and take it back one period it fits perfectly, however i cannot find and explanation for this.
so
does anyone know why it might be giving me a series 10 times smaller? (if i add standard deviations (sup and inf) in each direction these fit the series perfectly)
why is my forecast one period behind?
when i put forecast(e) the estimation box doesn't appear even though the help section says this is the correct option specification
thanks!
The program runs, however the result is around 10 times smaller than it should. It should give me a value in a range of -0.25 up to 0.25, however my forecast oscillates around -0.025 and 0.025.
When i multiply the forecasted series times 8 and take it back one period it fits perfectly, however i cannot find and explanation for this.
Code: Select all
_egarch.fit(e) f2 colcap_err2
series f20=f2*8 'this is the forecast that fits the real one
series sup2 = f2 + 1.96*colcap_err2
series inf2 = f2 - 1.96*colcap_err2
smpl 10/10/2009 @last
plot f2 d_colcap sup2 inf2
plot f20 d_colcap(-1) does anyone know why it might be giving me a series 10 times smaller? (if i add standard deviations (sup and inf) in each direction these fit the series perfectly)
why is my forecast one period behind?
when i put forecast(e) the estimation box doesn't appear even though the help section says this is the correct option specification
thanks!