Timeline for Continuation of GARCH(1,1) without data
Current License: CC BY-SA 4.0
13 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jul 24 at 20:16 | comment | added | dNyrM | Awesome! Thanks a lot. | |
| Jul 24 at 20:16 | vote | accept | dNyrM | ||
| Jul 24 at 19:35 | comment | added | Pleb | Yes, that's correct. | |
| Jul 24 at 19:01 | comment | added | dNyrM |
Cool! I see it. So old_resid comes from resid = returns - mu and old_conditional from conditional[t] where t is 2021-06-24? (from my code)? Did I get it right? Last question, I promise haha
|
|
| Jul 24 at 15:02 | comment | added | Pleb | Alright I have edited the code. Everything else still stands. Please see if this is what you desire. Otherwise feel free to provide feedback :-) | |
| Jul 24 at 15:01 | history | edited | Pleb | CC BY-SA 4.0 |
Edited to accommodate for authors request.
|
| Jul 24 at 14:31 | comment | added | dNyrM | Yeah, I'd have them. Re-accomodate and we're done here. Thanks a lot for your help. | |
| Jul 24 at 14:30 | comment | added | dNyrM |
Yeah, I'd have them stored somewhere. So the residual is resid and the vol estimate isconditional[t] where t is 2021-06-24? If so, could you fix your code to represent this? You could call it old_resid and such to express the placeholder state of the variable, since the old data would be unknown in this context. I'd award the points once it's done.
|
|
| Jul 24 at 14:27 | comment | added | Pleb | If you have these, I will redo the code to accommodate. Otherwise, I think you are out of luck since you cannot properly initialize the filter equation (last part of your code). | |
| Jul 24 at 14:17 | comment | added | Pleb | AFAIK: You want to use the already estimated GARCH parameters on new data, correct? In order for the GARCH process to run (eg. getting your estimate at time 2021-06-25 and after), you need the previous value of the volatility estimate and the residual in order to start the GARCH process out of sample. Hence, you need respectively the residual and the volatility estimate of 2021-06-24. If these are stored somewhere, you can use these. Otherwise, the process do not know at what "vol level" to begin at. | |
| Jul 24 at 13:50 | comment | added | dNyrM | Planning on doing a rolling window recalibration of parameters in N amount of time (perhaps monthly). Taking that into account, gotta ask something: you added the additional price values to the initial raw data and calculated the volatility using that and the predetermined GARCH parameters. What I'm looking, at this moment, is using only the predetermined GARCH parameters + the additional price values to calculate the conditional volatility for those days only, so any data that falls outside those price values wouldn't be available/used. Is this possible? Would your code need to be changed? | |
| Jul 24 at 13:46 | history | edited | Pleb | CC BY-SA 4.0 |
Added an important point.
|
| Jul 24 at 13:25 | history | answered | Pleb | CC BY-SA 4.0 |