Really brilliant video, but would you mind teach me the step about how to combine the histogram chart and VaR %loss in one chart? (the one you showed on end of video)
Tks a lot, great Tuto. I watched 4 videos before yours and I was still not confortable with the explanations and I work in trading (though we don’t compute the bar ourselves). Tks!
Thank you sir for your video. I have a question I am a finance student, and I am trying to calculate the CoVaR( Conditional value at risk) and VaR(Value at risk) using quantile regression, in order to analyze systemic risk for the banking system. So I already computed the coefficients alpha and beta for the CoVaR equation, using quantile regression in Eviews. Now i have to estimate the VaR for every bank when p=0.05 , So according to this approach, the VaR is equal to the total of quantiles computed for p=0.05 , using Rankit-cleveland definition. I should get a result that look like this, but I do not know how to do it : Descriptive Statistics for RATJ Categorized by values of RATJ Date: 11/13/17 Time: 00:57 Sample: 1/05/2010 11/03/2017 Included observations: 1956 RATJ Quant.* Obs. [-0.1, -0.05) NA 4 [-0.05, 0) -0.023874 816 [0, 0.05) 0.000000 1132 [0.05, 0.1) NA 4 All -0.017742 1956 *Quantiles computed for p=0.05, using the Rankit-cleveland definition. RATJ is the time serie for daily stock returns of the bank ATJ. Thank you so much for your time sir.
Thank you for this. This was a great video. I have a question though that, I believe,would help me a lot understand the whole VaR HS method. What date do I pick if I want to find today’s VaR? Also, are the prices from today or historic too?
Good question! Not too sure. But I suppose you could use some accounting measure like operating margin, with data generated over several quarters. You'd be calculating the worst case quarterly operating profit margin.
@@PatObi Thanks for your quick reply. The issue is that I work with biotech companies that are pre revenue. That implies the risks that they have are scientific, management , regulatory etc. I am researching on how I can build models like VAR but no success yet. Thanks
Prof... great video... I already calculated portfolio VaR using mean-variance method (Markowitz)... may I ask you some question pls ? 1) is it okay if I use the past 30 days portfolio returns (Jan 2017) to calculate the historical 1 day VaR ? 2) would you mind giving tutorial about VaR and time-varying volatility in excel , pls pls ? Thx Prof...
Pat, What would be good is a volatility adjustment to the historical simulation using EWMA and GARCH...... Contemporary performance embellishments should also be examined. Universities in my opinion fail to adequately equip students and merely re-teach 'old' tried and true methods. Portfolio optimization as taught in finance schools is a favorite gripe of mine. Best regards, Robert.
Tawfiq Bastaki : Thanks for your comment. The process for obtaining the frequency distribution data in Col K is briefly explained on about the 4:46th min of the video. On Excel, click as follows: Data => Data Analysis => Histogram => OK => for Input Range, highlight all returns on Col I => do nothing on Bin Range =>check labels if included => check Output Range and click on spreadsheet where output is desired => OK
very helpful!!! I have a question, my data is s&p500 and KOSPI two market, but the when i calculate the return some of the data can not be recognized , what can I do it? thanks
+yipan yu Hi Yu, make sure your data is all numeric. Yahoo!Finance is a good source for market data. In the Quote Lookup box, type in the symbol for Kospi (^KS11). then select Historical Prices on the left panel. Select dates and Get Prices. Follow my video instruction to calculate logarithmic return as follows =ln(Pt/Pt-1)
Hi pat thanks for sharing the valuable Video just a Quick one can we simple calculate Returns by =(T/T1)-1? I.E ( Current Price /Previous Price )-1 as output looks same for both
That's the discrete 'compounding' method of returns calculation. For empirical analysis, it's typical to use the logarithmic form (as demonstrated in the video), because it assumes continuous compounding.
That's because the assumption that you make when you calculate VaR is that the returns have a normal distribution. Log returns of assets in long period of times approximate to a normal distribution, arithmetic returns not necessarily
+Phillip Schumacher: Hi Phillip, apologies for delayed response. The values on spreadsheet are formatted to 2 decimal places. That rate of return, corrected to 4 decimal places, is equal to 0.2873%, calculated as ln(195.2/194.64).
simple, good, effective and practical! Awesome, thank you Pat Obi!
Hi, how did you add the actual VaR line on the distribution?
Love your work man! You saved tons of hassle for me hehe
Well simplified. Thanks Prof Pat
Thank you Professor, I have a question: for Variance covariance VaR can we use the LN returns ??
Really brilliant video, but would you mind teach me the step about how to combine the histogram chart and VaR %loss in one chart? (the one you showed on end of video)
Excellent Video! Thank you very much!
Hello, why do you use the lN formula to calculate the returns instead of using just the variation? thanks
Tks a lot, great Tuto. I watched 4 videos before yours and I was still not confortable with the explanations and I work in trading (though we don’t compute the bar ourselves). Tks!
best teacher your style is excellent very clear simple
A really solid explanation and demonstration.
Well done and regards!
thanks
Great Work.. you explained it well. My question is how did you get the $Return and frequency to be able to plot the graph?
Oh and by the way....
Merry Christmas to you and yours.
Thanks. Same to you!
Thank you sir for your video. I have a question
I am a finance student, and I am trying to calculate the CoVaR( Conditional value at risk) and VaR(Value at risk) using quantile regression, in order to analyze systemic risk for the banking system. So I already computed the coefficients alpha and beta for the CoVaR equation, using quantile regression in Eviews.
Now i have to estimate the VaR for every bank when p=0.05 , So according to this approach, the VaR is equal to the total of quantiles computed for p=0.05 , using Rankit-cleveland definition.
I should get a result that look like this, but I do not know how to do it :
Descriptive Statistics for RATJ
Categorized by values of RATJ
Date: 11/13/17 Time: 00:57
Sample: 1/05/2010 11/03/2017
Included observations: 1956
RATJ Quant.* Obs.
[-0.1, -0.05) NA 4
[-0.05, 0) -0.023874 816
[0, 0.05) 0.000000 1132
[0.05, 0.1) NA 4
All -0.017742 1956
*Quantiles computed for p=0.05, using the Rankit-cleveland definition.
RATJ is the time serie for daily stock returns of the bank ATJ.
Thank you so much for your time sir.
thank you very much my guy!
Thank you so much. Very helpful.
Very helpful! Brilliant work!
+Wong Dingyao Thanks!
Pat, great video and easy to understand. Do you know how to calculate Futures VaR? I do have a projct realting to it that I am having some issues.
I'll look into it and will post an update once I figure it out.
enjoyed the level of detail for steps thanks
Thank you for this. This was a great video. I have a question though that, I believe,would help me a lot understand the whole VaR HS method. What date do I pick if I want to find today’s VaR? Also, are the prices from today or historic too?
With historical DAILY data, it would be the VaR you calculate. Refer to the last few minutes of the video.
Excellent Pat. Is there anyway to analyse risks for private companies. For public we do have these prices, but for private, we don't. Any pointers?
Good question! Not too sure. But I suppose you could use some accounting measure like operating margin, with data generated over several quarters. You'd be calculating the worst case quarterly operating profit margin.
@@PatObi Thanks for your quick reply. The issue is that I work with biotech companies that are pre revenue. That implies the risks that they have are scientific, management , regulatory etc. I am researching on how I can build models like VAR but no success yet. Thanks
good explanation
I think you need to add quantity and market value on each day instead of multiplying the return to base market value when you started ?
Brilliant!
+Gus Montano Thanks.
Prof... great video... I already calculated portfolio VaR using mean-variance method (Markowitz)... may I ask you some question pls ?
1) is it okay if I use the past 30 days portfolio returns (Jan 2017) to calculate the historical 1 day VaR ?
2) would you mind giving tutorial about VaR and time-varying volatility in excel , pls pls ?
Thx Prof...
There is a special place for you in heaven! Thanks bro
Very kind. Thanks.
Nice explanation
Thank you
thak you. that was really helpful for me.
Thank you ❤
Pat,
What would be good is a volatility adjustment to the historical simulation using EWMA and GARCH......
Contemporary performance embellishments should also be examined. Universities in my opinion fail to adequately equip students and merely re-teach 'old' tried and true methods.
Portfolio optimization as taught in finance schools is a favorite gripe of mine.
Best regards,
Robert.
Thanks Robert. Great suggestion.
Many thanks 🙏
Is there a book/working paper that describes these steps that I could cite?
Hi,
Did you find one?
Thank you very much
very interesting but could you please explain the figures on Column K and how to calculate them.
Tawfiq Bastaki : Thanks for your comment. The process for obtaining the frequency distribution data in Col K is briefly explained on about the 4:46th min of the video. On Excel, click as follows: Data => Data Analysis => Histogram => OK => for Input Range, highlight all returns on Col I => do nothing on Bin Range =>check labels if included => check Output Range and click on spreadsheet where output is desired => OK
very helpful!!! I have a question, my data is s&p500 and KOSPI two market, but the when i calculate the return some of the data can not be recognized , what can I do it? thanks
+yipan yu Hi Yu, make sure your data is all numeric. Yahoo!Finance is a good source for market data. In the Quote Lookup box, type in the symbol for Kospi (^KS11). then select Historical Prices on the left panel. Select dates and Get Prices. Follow my video instruction to calculate logarithmic return as follows =ln(Pt/Pt-1)
Hi pat thanks for sharing the valuable Video just a Quick one can we simple calculate Returns by =(T/T1)-1? I.E ( Current Price /Previous Price )-1 as output looks same for both
That's the discrete 'compounding' method of returns calculation. For empirical analysis, it's typical to use the logarithmic form (as demonstrated in the video), because it assumes continuous compounding.
Should we use LN to find the returns.. or (latest price - previous price )/ previous price ..?
That's because the assumption that you make when you calculate VaR is that the returns have a normal distribution. Log returns of assets in long period of times approximate to a normal distribution, arithmetic returns not necessarily
thanks alot , excellent, can you please share with me your excel sheet ?
may i know how the portfolio is calculated
Simply add the $ returns of the two assets....4th minute of the video.
Great video thank's but if I multipy 700.000$ with 0,29% I get 2.030$ (min.: 03:09) I'm getting a little confused
+Phillip Schumacher: Hi Phillip, apologies for delayed response. The values on spreadsheet are formatted to 2 decimal places. That rate of return, corrected to 4 decimal places, is equal to 0.2873%, calculated as ln(195.2/194.64).
Hi, may i know how did you get the red line for the 99% confidence level
+Jay Teo It's the 1st percentile of the distribution, calculated as =PERCENTILE(array of cells containing the returns, 1%)
To plot the histogram * I mean
Thank you Professor, I have a question: for Variance covariance VaR can we use the LN returns ??
Yes.