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
@@user-wg7nw3mh2e It is called "Autoregressive model". Just in case, here is the link: docs.google.com/spreadsheets/d/1wMRRODv3W3qDbdbE1vn3QFioDNLKQEy1/edit?usp=sharing&ouid=113436662715404606257&rtpof=true&sd=true
Hey man great!!!!! Like the haircut lol. This is one of the most confusing and basic models in Finance and often folks take so much time to make any sense of it. Great work on the explanation and Excel!
What would be absolutely gorgeous and what is until now not available on youtube: Please add another vid on how to implement an ARMA-Model. And please compare it with the basic AR and MA approach and please show us how to select betweent these models based on predicting power. Please explain how to compute ACF and PACF with excel so that one could choose how many Lags to use for these models. This would be so great. I have read about all these things but I do not how to do it and how to implement it. Please. On the day I see these vid on your channel, I will become your biggest fan and supporter :-)
Hi Grimm, and thanks for the question! This might be a topic I will cover someday, however as ARMA is just a combination of AR and MA, and ARIMA is effectively the same with first/second/etc. differencing, most of the relevant concepts I feel are already covered in the existing tutorials :)
Another amazing video Savva! I'll be testing some my models based on your videos. So, I would like to ask you: 1. what are we doing if let's say 8 out of 10 regression coefficients are statistically insignificant? 2. What if we find after calculating polynomial roots that the model is instable? How do we correct and improve it? Thanks for your amazing creativity!
Hi Ivan, and many thanks for the excellent questions and your contributions to the channel! If the results are unclear from the individual coefficients, you can always use an F-test for the significance of the model overall (for example, if you run an AR(10), and only the third and the sixth lags are significant and you cannot have an a priori reason to suspect they are somehow special, you simply test for the significance of the R-squared). If the model is unstable, it is generally advisable to apply any sort of autoregressive modelling to the first differences (or other similar transformations) of the variable. For example, if an autoregressive model of inflation is unstable as per the inverse roots of characteristic polynomial test, you can apply autoregressive modelling to differences in inflation.
So precise explanation that makes it look so simple. Thank you Sava!! As for future suggestions, what's your opinion on Michaud portfolio optimization method?
Hi Peter, and thanks so much for the kind words! Michaud resampled efficient frontiers are an interesting idea that definitely warrants investigation. I am currently figuring out how to best condense the model into a tutorial and will definitely release a video on these in the future.
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
I don't see this spread sheet on the drive
@@user-wg7nw3mh2e It is called "Autoregressive model". Just in case, here is the link: docs.google.com/spreadsheets/d/1wMRRODv3W3qDbdbE1vn3QFioDNLKQEy1/edit?usp=sharing&ouid=113436662715404606257&rtpof=true&sd=true
@@NEDLeducation sorry. I found it. I tried to delete this comment but thank you very much for the quick reply.
thank you, I am really need this today
All your video are excellent!
Excellent work and exposition!
Hey man great!!!!! Like the haircut lol. This is one of the most confusing and basic models in Finance and often folks take so much time to make any sense of it. Great work on the explanation and Excel!
Thanks so much for the kind words! I plan to release even more videos on foundational time series econometrics in the near future, so stay tuned! :)
Pretty cool way of looking at the characteristic polynomial..awesome
What would be absolutely gorgeous and what is until now not available on youtube: Please add another vid on how to implement an ARMA-Model. And please compare it with the basic AR and MA approach and please show us how to select betweent these models based on predicting power. Please explain how to compute ACF and PACF with excel so that one could choose how many Lags to use for these models. This would be so great. I have read about all these things but I do not how to do it and how to implement it. Please. On the day I see these vid on your channel, I will become your biggest fan and supporter :-)
Hi, and thanks for the feedback, really appreciate it! I do plan on covering these key concepts in time series modelling very soon, so stay tuned!
great work
Hello, you have a video on the AR and the MA models but not on the ARMA and ARIMA models. Could be something
Hi Grimm, and thanks for the question! This might be a topic I will cover someday, however as ARMA is just a combination of AR and MA, and ARIMA is effectively the same with first/second/etc. differencing, most of the relevant concepts I feel are already covered in the existing tutorials :)
Another amazing video Savva! I'll be testing some my models based on your videos. So, I would like to ask you: 1. what are we doing if let's say 8 out of 10 regression coefficients are statistically insignificant? 2. What if we find after calculating polynomial roots that the model is instable? How do we correct and improve it? Thanks for your amazing creativity!
Hi Ivan, and many thanks for the excellent questions and your contributions to the channel! If the results are unclear from the individual coefficients, you can always use an F-test for the significance of the model overall (for example, if you run an AR(10), and only the third and the sixth lags are significant and you cannot have an a priori reason to suspect they are somehow special, you simply test for the significance of the R-squared). If the model is unstable, it is generally advisable to apply any sort of autoregressive modelling to the first differences (or other similar transformations) of the variable. For example, if an autoregressive model of inflation is unstable as per the inverse roots of characteristic polynomial test, you can apply autoregressive modelling to differences in inflation.
So precise explanation that makes it look so simple. Thank you Sava!! As for future suggestions, what's your opinion on Michaud portfolio optimization method?
Hi Peter, and thanks so much for the kind words! Michaud resampled efficient frontiers are an interesting idea that definitely warrants investigation. I am currently figuring out how to best condense the model into a tutorial and will definitely release a video on these in the future.
@@NEDLeducation Thank you so much Sava! I believe a video of yours with this method would be really interesting and helpful!
THis is amazing!!
Why are the degrees of freedom n-10-9-1? Shouldn't it be only n-9-1?
What is the name of the formulars? Refer to 14.35 of the video please.
Hi, and thanks for the question! This is the characteristic polynomial formula for the autoregressive equation.