You can find the spreadsheets for this video and some additional materials here: drive.google.com/drive/folders/1sP40IW0p0w5IETCgo464uhDFfdyR6rh7 Please consider supporting NEDL on Patreon: www.patreon.com/NEDLeducation
Please help me can you help me to calculate the forecasting result from arima(0, 1, 2) egarch(0, 4) model? which data should be included in this model, and how to find forecasting results manually? and adjust it with the output of eviews. I use stock data then convert it to return data
Hi, and thanks for the question! For ARIMA(0,1,2) you will need to calculate the first differences of the time series (I = 1), do not include any lags (AR = 0), and include two lagged residual terms in the estimation (MA = 2). You could fit it iteratively by applying OLS and inputting residuals from the past iteration to the next one until convergence or use maximum likelihood. Not sure what you are referring to with EGARCH(0,4) though, as it does not make sense to include no immediate disturbance terms in the variance equation (you need at least one) alongside four lags of conditional volatility.
Hi Vaibhav, excellent question! This is due to the fact that immediate disturbance terms are scaled by conditional volatility and that the threshold (or, as it is also called, leverage) term theta is included. This is one of the reasons EGARCH can be more flexible than more standard GARCH approaches. Hope it helps!
@@NEDLeducation Hi, thanks a lot for the video:). But This explanation is still a bit confusing to me. How is immediate disturbance terms scaling by conditional volatility and not involving alpha in the long run volatility related?
Hi Andy, and thanks for the question, I have got a video on alternative distribution specifications in GARCH, check it out if you are interested: ruclips.net/video/pwGXftsrWYE/видео.html. Obviously, you can use the same logic to perform EGARCH or any other GARCH parametrisation by simply changing the probability density function in log-likelihood calculations. Hope it helps!
You can find the spreadsheets for this video and some additional materials here: drive.google.com/drive/folders/1sP40IW0p0w5IETCgo464uhDFfdyR6rh7
Please consider supporting NEDL on Patreon: www.patreon.com/NEDLeducation
Excellent and fabulous demo! thanks a lot.
What if, in terms of interpretation, only alpha is statistically significant while theta is not ?
Seems like T-GARCH better fit to upper end of VIX compared to EGARCH, hence better practical simulation :)
Great job; thanks :)
Thank you for this so much. Could you please show us how we can do EGARCH in Python? Will be much appreciated.
Please help me
can you help me to calculate the forecasting result from arima(0, 1, 2) egarch(0, 4) model?
which data should be included in this model, and how to find forecasting results manually?
and adjust it with the output of eviews.
I use stock data then convert it to return data
Hi, and thanks for the question! For ARIMA(0,1,2) you will need to calculate the first differences of the time series (I = 1), do not include any lags (AR = 0), and include two lagged residual terms in the estimation (MA = 2). You could fit it iteratively by applying OLS and inputting residuals from the past iteration to the next one until convergence or use maximum likelihood. Not sure what you are referring to with EGARCH(0,4) though, as it does not make sense to include no immediate disturbance terms in the variance equation (you need at least one) alongside four lags of conditional volatility.
@@NEDLeducation i think i need send you picture to explain my question, may i know and send it to your email address or sosial media account?
Hello, excelent, clear and direct. In likewood function Vt*sqrt(2*pi) ¿should be vt^2*sqrt(2*pi)?
my mistake, you calculate the square vt into the sqrt and present out the sqrt
Hi Carlos, and thanks for the comment! Glad you enjoyed the video!
why in logn run volatility we do not reduce alpa from 1 and only do beta?
Hi Vaibhav, excellent question! This is due to the fact that immediate disturbance terms are scaled by conditional volatility and that the threshold (or, as it is also called, leverage) term theta is included. This is one of the reasons EGARCH can be more flexible than more standard GARCH approaches. Hope it helps!
@@NEDLeducation Hi, thanks a lot for the video:). But This explanation is still a bit confusing to me. How is immediate disturbance terms scaling by conditional volatility and not involving alpha in the long run volatility related?
if the residue obey the t distribution, how to handle it?
Hi Andy, and thanks for the question, I have got a video on alternative distribution specifications in GARCH, check it out if you are interested: ruclips.net/video/pwGXftsrWYE/видео.html. Obviously, you can use the same logic to perform EGARCH or any other GARCH parametrisation by simply changing the probability density function in log-likelihood calculations. Hope it helps!