DCC GARCH model: Multivariate variance persistence (Excel) 

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

This is amazing - I would love to see the DCC application with 3 or 4 assets :D :D :D

I would really appreciate if you make a video of DCC Garch Model of 3 assets, I just found your channel while struggling to applicate this model in my thesis and I have to say that you are doing an amazing job. If you find time please make the video. thanks!

You rocked it !! so much appreciate for your work

Thank you for this

Eager to see how this model works with multiple assets (3 for instance) in terms of volatility spillover effects and hedging. Thanks for the effort!

Great tutorial. Thanks for the explanation.

Thank you very much for your Video! Best video for DCC GARCH ever!

This is the best DCC-GARCH video out there. Goes into the mathematics without drowning you in detail, and was super helpful to wrap my head around DCC-GARCH's workings for my MSc Finance dissertation. Many thanks!

You did a very good job!

thank you so much for the tutorial! Since you asked this at 10:40, I would love it if you perform a DCC with multiple assets

Thank you so much for the video, this is the most useful and practical explanation of DCC GARCH on YT. Video suggestion: BEKK model, CCC GARCH and OGARCH.

Thanks you for the very helpful video. The iterative equation for Q_t in the spreadsheet appears to differ from the original DCC model of Engle in two non-trivial aspects (see, e.g., "Theoretical and Empirical properties of Dynamic Conditional Correlation Multivariate GARCH"). First, the spreadsheet uses residuals (epsilon_t) on the right-hand side of the iterative equation, while the original DCC uses standardized residuals (epsilon_t over conditional volatility estimated from univariate GARCHs). Second, the original DCC defines Q_bar as the unconditional covariance of standardized residuals resulting from univariate GARCH estimation, but the spreadsheet uses parameters of univariate GARCHs and a long-run correlation parameter to construct the Q_bar matrix. Can you please explain these differences and whether your implementation is the same as the model proposed in the original paper? Thanks again!

amazing tutorial. :)

Hi NEDL, that's the best tutorial on GARCH DCC I've ever seen! I'd love to see the model performence with multiple assets! Is there maybe any video of yours, containing this issue or are You planning to make one?

Sir NEDL thx for all the videos, they are amazing, could you do the dcc with more assets and using egarch? thx again!! love the channel!

I just decided to become your patron after watching this AWSOME VIDEO. Unbelievble what you have done. Wish you were my University Professor. I support peoples suggestion about explaining Academic papers. Other suggestions for videos 1. 3 asset DCC 2. COPULAS Thank you so much. You are a genius.

Thanks also kindly cover shoc transmission, volatility spillover when u cover dcc/bekk models in econometric packages like reviews or r or estima etc.

Amazing job again mate! Keep it up! You have a bright future ahead! Small suggestion for future videos: Try recreating financial academic papers (like "Asset Pricing Models and the Norwegian Stock Market" by Eivind Rossvoll from NTNU).

Hi from Brazil, I would like to estimate a garch-Midas of volatility stocks return and add accounting variables to explain this volatility, have you ever seen this in eviews?

Hi Savva, thanks for your video, great content,. Just a questions, in your estimation process you have made the starting correlation to be estimated. In a expanded situation with more than 2 assets, the correlation will be a matrix, how do yo enforce it to be symetrical? Any suggestions? thanks

what an excel, ive been playing with it. this a hidden gem!!!! Ive been reading some Papers on Using Garsh and DCC for finding correlation of Commodities like oil and oil companies, any experience with that?

Thanks for all the help! I have done all of this but expanded it to a set of 23 assets. The only thing that isn't working is that the final matrix Ht has a negative determinant, which is causing errors in the log-likelihood function. Do you have any idea what is causing this error?

LOVE IT, would love a multiple asset (3-4 assets +) DCC GARCH, and is it possible to create Discrete State Hidden Markov Models to find Bullish, Neutral and Bearish states based on the Garch Volatility. Thanks so much, love all your work!

Thanks a lot for the video. If you have a chance, please make a video with more than one asset. Thanks!

this is wonderful

Hi Savva, Is there any way to find dynamic hedge ratio in excel?

hello how did you calculate q bar and immediate distribution here. Is there any video i could watch for that? please help me out

Interesting model - Do you know the original that was published on the DCC GARCH? Is it Engle, Sheppard (2001) "Theoretical and empirical properties of dynamic conditional multivariate GARCH"?

Could u please explain common sense behind immediate disturbance coming from residuals and persitence of time series

Hello NEDL, do you have any literature recommendation for DCC GARCH?

Could you do an example with more securities or a different GARCH model?

Hello needle I am a bit confused on how you found the DCC for apple and s&P. It shows INDEX but Could you walk through how you calculated that?

hello ned, how can i get the probability of the ARCH AND GARCH parameters?

hi, do you have something like this for a BEKK model ?

What about DCC and CCC (Constant Conditional Correlation) in Python? It's easy to estimate them in E-Views and R programming. I've already calculated them in Excel. But it was a real madness! Can you estimate them in python and R?

Hi, i was studying "Is there any volatility spill over between oil price volatility and stock market returns? or Does oil price volaility move togather with stock market volatility? The DCC alpha value was insignificant but DCC beta value was significant. what does it means?

Great job as always, this is the best tutorial video about DCC GARCH model! I am trying to replicate this study but using 10 stocks instead but what I think it is really strange is that the parameters "a" and "b" (Dynamic Correlation coefficients) are the same for the 10 ETFs, the DCC model considers that all the 10 stocks have the same correlation dynamics? is there a way to consider different correlation dynamics? Thanks in advance and please keep doing this YT videos, they are great!

Is Q-bar a long-run correlation matrix calculated from long-run volatility in GARCH? Do I understand this right?

how to deal with non normal distribution in returns, say heavy tailed ?

Thank you for this great explanation. I have a short question: Do the matrix calculations in sheet 'DCC estimation' vary based on whether a Garch(1,1) or Garch (p,q) process is used for the conditional volatility. Or is the process the same for both? Thank you in advance!

Please do multiple assets dcc garch🙏

how did you find the Alpha and beta values ARCH (Alpha) and GARCH (Beta)? Thank you

Thank you for the tutorial. I have a question (may be a dumb question), but how come we don't add all the GARCH constraints here, like alpha+beta0, beta between 0 and 1, as you did in your GARCH video (and that video is just as wonderful as this one thanks!!!). I look forward to your reply :)

Hello, Im trying to using bitcoin and stocks for DCC Garch analysis, but I realise bitcoin have 7 days data and stocks only have 5 days data. What should I do?

The calculation of log likelihood in T16, you have multiplied n* n * ln (pi) instead of n* ln (pi).

Hi and thank you for this video! I applied this model and came up with constant correlation values for two assets. Then I dropped the condition that B11 and B12 must be greater than or equal to 0 and got dynamic correlation values. What does this mean for the model? Is this invalid or can I continue to use it as it is?

Hi Savva, first of all thank you so much for providing all those great videos, I think its a gift for every finance student! Currently I am doing a DCC-GARCH to interpret the safe haven properties (I saw your other vid as well) of cryptocurrencies at certain economic turmoil scenarios and I face the problem that for MSCI_World + ETH (2016-2021 period) the solver results in a dynamic correlation of a) 1.8e^-11 so very small and b) exactly 0. Log likelyhood 7412 and a starting correlation of 0.0245. How to interpret this especially the dynamic correlation or what can I add to change the result? Since DC b)=0, Qt/Q_Bar also does not change and Qt* and Rt neither. Ultimately resulting in a correlation unter DCC_result tap, which is totally wrong showing 0.0245/1.00 in a repetitive order row by row. How can I fix it? May it be the best idea to force DC b)>0? Is this even possible? Best PS: May this problem be due to the very strong growth of ETH since its ICO resulting in a long-run-volatility of over 7.X%

Many Thaks for the Video. Can you help me out in showing DCC with EGARCH? Thanks once again for this fantastic video.

Are the returns standardized?

Please explain bivariate BEKK model

Thanks buddy... can i share you a financial data series (2 variables) for making BEKK bivariate garch model for your future video?

DECO Model please ?