Greeting, it was wonderful explanation, could you please calculate the whole procedure of component VAR of multiple securities (atleast 4), it would be your kindness, regard
Hi, How is this different from calculating the portfolio returns, computing it's mean and st. deviation and then using them as inputs for the norm.inv function?
Hi there, First off, thanks for the work you have done, much appreciated. Could you by any chance send me a copy of the spreadsheet or post a link to it. I think it would be very helpful for those trying to work through this to work along side you as you explain things. thanks again John
Hi,Thanks for your work you shared.I have a question could you please tell me if we have student t distribution then how can we set our cutoff point. Best Regards
hey, thank you for uploading such a great explanation video. Anw, i have a question. I try to do the variance-covariance matrix steps for 1198 observations. but when i input the formula it says error value. do u think it can happens bcs theres missing data? thank you.
In your portfolio, you have taken all the equity stock prices. But what to do in case we have forwards, bonds ot traury bills, CFD's etc. in our portfolio. How will we make variance covariance matrix then? Do we need to treat the other asset classes also in the same way as for equity????
+Manish Pratap Singh Hi. That's a good question. As you can see in the video we only work with the returns on the assets. If it is possible to standardize every asset in the portfolio and if you can get returns on those assets then you can apply the variance covariance matrix to any asset class.
Hello, Thanks for this video it was very helpful. But i have a question : You calculate the portfolio volatility based on 17-Days data and you used this volatility to determine 1-Day VaR, don't you have to use the 1Day volatility to compute 1Day VaR ? Thanks
Hi. Good question! Volatility based on a single day of training may not provide the most accurate result for 1 day VaR, as prices may change a lot one day and very little the next day. Also, We often cannot obtain asset prices During a day of trading as in most cases asset prices are recorded at the end of the trading day. Therefore, volatility over a month may provide the best representation of asset price movement for a 1 day VaR estimation. Hopefully this helps.
also, you have 18 data points but you divide by 17 at time=3:11 of the vid. is there are reason for that? are you actually using n-1 or is there a mistake in the formula? thanks again, John
Thank you very much for spotting this John. You are right, you should divide by 18 observations not 17. The procedure remains the same but the answer will change a little bit.
Hello, thank you so much for your video it was very helpful. Just one thing to note: you should actually divide by (n-1) not by n, so you were right dividing by 17, not 18; even though you made a mistake =P. If you don't divide by 17, the variances in the Covar matrix will not be the same for the same stock (variance between AMZN and AMZN)
For a long-short portfolio you usually trade spreads. So you have a ratio with the long asset on the numerator and the short asset on the denominator. So instead of taking the returns of each asset separately you calculate the returns of the whole ratio. In other words you treat the entire ratio as a single asset and you always want the ratio to go up in order to have a positive return. So in conclusion you calculate VaR using the returns and the standard deviation of the long-short ratios instead of separate assets.
+Sania K Hi. When dealing with VaR, long time periods can lead to imprecise results. The example uses a 10 day Var at 1% probability of loss, which in general can be interpreted as seeing VaR loss approximately once in 3 years (100*10=3 years). Also, as per Basel Regulatory Framework, financial institutions (banks) are required to estimate VaR at 99% confidence level (1% probability of loss) for 10 business days. While the 10 day period is the standard procedure, different portfolios can demand different time horizons and the same formula can be applied. Hopefully this helps.
+claussy20ten Thank you so much for your reply! It was very helpful :) I was also looking in to the monte carlo simulation to calculate the VAR. I was hoping to use t-distribution to do this but i cant find anything(specific to excel) helpful online. Do you have any references that u believe might be helpful?
It was very useful. Thank you so much!!!
Greeting, it was wonderful explanation, could you please calculate the whole procedure of component VAR of multiple securities (atleast 4), it would be your kindness, regard
Great content!! Thanks a lot buddy!
Hi,
How is this different from calculating the portfolio returns, computing it's mean and st. deviation and then using them as inputs for the norm.inv function?
MMULT (TRANSPOSE ) DATA MUST BE DIVIDED BY N-1
With such a small data set, treating the data as a population instead of a sample size should make very little difference
Hi there,
First off, thanks for the work you have done, much appreciated.
Could you by any chance send me a copy of the spreadsheet or post a link to it. I think it would be very helpful for those trying to work through this to work along side you as you explain things.
thanks again
John
Hi, is VaR at 1% level equal to confidence level of 99%?
I didnt found any example with multi currencies + bonds portfolios to find the VAR. Please help !
quite not clear why we have subtracted cuttoff from initial investment ????
i mean looking at the formula cuttoff itself looks like VaR
Hi,Thanks for your work you shared.I have a question could you please tell me if we have student t distribution then how can we set our cutoff point.
Best Regards
Hi, thank for your video!!
Did you use the asset-normal method or delta normal method??
Thank you!!
I keep getting the value error when I try work out the portfolio signa.
hi, could you send me this .xls sheet for learning purposes?
hey, thank you for uploading such a great explanation video. Anw, i have a question. I try to do the variance-covariance matrix steps for 1198 observations. but when i input the formula it says error value. do u think it can happens bcs theres missing data? thank you.
CTRL + SHIFT + ENTER
In your portfolio, you have taken all the equity stock prices. But what to do in case we have forwards, bonds ot traury bills, CFD's etc. in our portfolio. How will we make variance covariance matrix then? Do we need to treat the other asset classes also in the same way as for equity????
+Manish Pratap Singh
Hi. That's a good question. As you can see in the video we only work with the returns on the assets. If it is possible to standardize every asset in the portfolio and if you can get returns on those assets then you can apply the variance covariance matrix to any asset class.
Someone has the spreadsheet for download?
Hello, Thanks for this video it was very helpful.
But i have a question : You calculate the portfolio volatility based on 17-Days data and you used this volatility to determine 1-Day VaR, don't you have to use the 1Day volatility to compute 1Day VaR ?
Thanks
Hi. Good question! Volatility based on a single day of training may not provide the most accurate result for 1 day VaR, as prices may change a lot one day and very little the next day. Also, We often cannot obtain asset prices During a day of trading as in most cases asset prices are recorded at the end of the trading day. Therefore, volatility over a month may provide the best representation of asset price movement for a 1 day VaR estimation. Hopefully this helps.
Shouldn’t you divide by 16 instead 17 ?
Wait I just realized you have 18 inputs total. Good tutorial, man.
when applied the portofolio sigma formula's the result error as #num can you help me to fix it?
Hi, thank you for the video.
Why do you use a lognormal function to calculate the returns?
Hi. When using ln function we assume that asset prices are time additive. The ln returns are also assumed to be normally distributed.
+claussy20ten Perfect, thank you very much for your fast answer!
also, you have 18 data points but you divide by 17 at time=3:11 of the vid.
is there are reason for that? are you actually using n-1 or is there a mistake in the formula?
thanks again, John
Thank you very much for spotting this John. You are right, you should divide by 18 observations not 17. The procedure remains the same but the answer will change a little bit.
hey, but when we take ln(Pt+1/Pt), there will be only 17 samples instead of 18 right?
Hello, thank you so much for your video it was very helpful. Just one thing to note: you should actually divide by (n-1) not by n, so you were right dividing by 17, not 18; even though you made a mistake =P. If you don't divide by 17, the variances in the Covar matrix will not be the same for the same stock (variance between AMZN and AMZN)
Someone has this sheet to download ?
Hello have you tried to calculate a VaR for long and short positions portfolio?
For a long-short portfolio you usually trade spreads. So you have a ratio with the long asset on the numerator and the short asset on the denominator. So instead of taking the returns of each asset separately you calculate the returns of the whole ratio. In other words you treat the entire ratio as a single asset and you always want the ratio to go up in order to have a positive return. So in conclusion you calculate VaR using the returns and the standard deviation of the long-short ratios instead of separate assets.
Great video! But one question: what if we had short positions instead?
I have made the same question....since i cannot work this out ..
Hi 👋 did you get a response for this?…trying to apply a similar concept for FX Portfolio VaR and wondering what to do with short positions….
hi, thank you so much for posting this! I just wanted to ask if it was possible to calculate a 10 year var using this method?
+Sania K Hi. When dealing with VaR, long time periods can lead to imprecise results. The example uses a 10 day Var at 1% probability of loss, which in general can be interpreted as seeing VaR loss approximately once in 3 years (100*10=3 years). Also, as per Basel Regulatory Framework, financial institutions (banks) are required to estimate VaR at 99% confidence level (1% probability of loss) for 10 business days. While the 10 day period is the standard procedure, different portfolios can demand different time horizons and the same formula can be applied. Hopefully this helps.
+claussy20ten Thank you so much for your reply! It was very helpful :) I was also looking in to the monte carlo simulation to calculate the VAR. I was hoping to use t-distribution to do this but i cant find anything(specific to excel) helpful online. Do you have any references that u believe might be helpful?
PORFOLIO PROPORTIONS ,pls?