Very very good video, can you please make play lists and number the topics in an order, you are very good teacher and I think you channel can be very good if you just order it a little bit. Good luck.
This is a very good explanation and presentation. I have a question regarding diagnostics: What types of diagnostics should we use for the model? Should we check the standard diagnostics for the each univariate GARCH model(autocorrelation, (conditional)heteroskedasticity, model stability, residual distribution fit) and also what kinds of diagnostics for the MGARCH estimation? Thank you!
alpha is a parameter you're trying to estimate. Numerical algorithms will take a starting value, like alpha = .5, compute the likelihood, and travel in a direction towards greater likelihood values.
Dear Sir Thank you so much for this video I need to construct a News impact curve in my Ph.D work. I checked every were to get some idea about the same Only one video that i got is yours Still i have some confusions Will you pls assist me to do the same Best Regards Neenu It will be beneficial for me Please do respond to the query
It is considered in Hamilton's 1994 textbook "Time Series Analysis". It is implemented in PcGive for OxMetrics, and I bet that there exist packages for R, Python, etc. as well.
For a given number of observations, the expectation/mean of the estimator does not equal the true (population) value of \phi (in general). The estimator is asymptotically unbiased, that is, consistent (under suitable conditions).
thank you so much for the vedio. By the way, what is the correlation here actually describing? the corr between return? or the corr between volatility>
Hello Mr. Pedersen, thank you for this video, very helpful. But please can you assist me with the link video on how I can run a DECO GARCH model with OXMETRICS. Thank you
slide 28/45, slite mistake, (~z)t-1=(1,zt-1',zt-2',...) and size is (1+pk)*1 instead of (~z)t-1=(1',zt-1',zt-2',...). otw great explanation, thank you a lot
Indeed you can estimate the GARCH model parameters by GMM instead of MLE. The idea is simply to state relevant moment conditions. See e.g. swopec.hhs.se/hastef/papers/hastef0434.pdf