Corret me if I am wrong: With the non-parametric ES we simply average the returns that are lower than our desired percentile (e.g the lower 5% of last year's returns) - ok. My question is, why do we use average? Isn't this bound to be affected by extremely rare but negative events? Why don't we use a weighted average? Or a median? Many thanks
Hi Frosty, short answer: yes, we can (for non-param/discrete distributions also). ES is a special case of the a spectral measure (which is a special case of a general risk measure) where the tail losses are equally weighted, however a spectral measure has "weakly increasing" (ie, not decreasing) weights so it's more natural state is increasing weights as losses are greater.
At around 3:53, the Y-Axis (which represents probability) is showing data corresponding to 10% of probability, NOT 5% as claimed in the video, isn't it?
Hi Sriman, actually the red AREA under the curve is 5% per =NORM.S.INV(5.0%) = -1.644853627 or rounded -1.645 and at the same time =NORM.S.DIST(-1.645, FALSE) = 0.10313564 such that you are correct in one sense: the pdf curve hit the point (-1.645, 0.10313564) but this is the density function. The 0.103 is not a probability; this is a continuous function. I hope that helps.
Hello friends, I have a few questions: 1 / Risks will be specified after we have identified the audience, objectives, and operational processes ?. 2 / Risk will be directly integrated into the business process ?. 3 / The Risk department is responsible for determining the VaR (Value at Risk) and presenting it to the Board of Directors seeing the risks and proactively preventing them? 4 / Actively preventing risks will help us improve the value of products / services to customers?
Hi. How it would be possible to calculate the VAR for a company which holds money in a bank account? It would make sense to assess the bank financial soundness, calculating the CAP ratio e than calculating the VAR for that company? what formula might be applied for this kind of calculation? thanks!
Hi thanks for the great video, I am doing thesis using GARCH-MIDAS model using Generalized Hyperbolic distribution, do you have any idea how to compute VAR in R using these models and distribution?
We select the confidence level. Typical selections are 95.0%, 99.0% or 99.0% because we are typically interested in levels of loss that shouldn't be exceeded except rarely; e.g., 99.0% confident VaR means "we expect this loss to be exceeded only 1.0% of the time."
My lecturer in a very expensive business school couldn’t explain that for 3 weeks... thank u so much man
Nowadays education is not a matter of money anymore but willing to learn
Yeah... My teacher taught this for hours but this video clearly demonstrates what it is in only 10 minutes!
I'm from Germany and your videos are great! Easy to understand. Keep it up !
Thank you for watching! We are so happy to hear that our videos are so helpful!
Great instructor of our modern era!, thank you so much for simplicity and practicality
god bless you. u explained what my book and lecturer couldnt
Thank you for watching! We are very happy to hear that our video was so helpful.
Nice explanation, clear, concise. Keep up your good work.
Great video very well explained
Thank you for watching! We are happy to hear that our video was helpful!
Corret me if I am wrong:
With the non-parametric ES we simply average the returns that are lower than our desired percentile (e.g the lower 5% of last year's returns) - ok.
My question is, why do we use average? Isn't this bound to be affected by extremely rare but negative events? Why don't we use a weighted average? Or a median?
Many thanks
Hi Frosty, short answer: yes, we can (for non-param/discrete distributions also). ES is a special case of the a spectral measure (which is a special case of a general risk measure) where the tail losses are equally weighted, however a spectral measure has "weakly increasing" (ie, not decreasing) weights so it's more natural state is increasing weights as losses are greater.
Great explanation, thank you
At around 3:53, the Y-Axis (which represents probability) is showing data corresponding to 10% of probability, NOT 5% as claimed in the video, isn't it?
Hi Sriman, actually the red AREA under the curve is 5% per =NORM.S.INV(5.0%) = -1.644853627 or rounded -1.645 and at the same time =NORM.S.DIST(-1.645, FALSE) = 0.10313564
such that you are correct in one sense: the pdf curve hit the point (-1.645, 0.10313564) but this is the density function. The 0.103 is not a probability; this is a continuous function. I hope that helps.
Why P(L > VaR) VaR) = 1-c
maximum loss that an investor can put up with; so that defines amount of collateralization after default.
Thanks a lot, very clear explanation for VAR.
Hello friends,
I have a few questions:
1 / Risks will be specified after we have identified the audience, objectives, and operational processes ?.
2 / Risk will be directly integrated into the business process ?.
3 / The Risk department is responsible for determining the VaR (Value at Risk) and presenting it to the Board of Directors seeing the risks and proactively preventing them?
4 / Actively preventing risks will help us improve the value of products / services to customers?
Whats the typical var of a market portfolio like s and p 500 ?
What's formula in the cell for VaR?
it is just similar to " p" value. isn't it?
Hi. How it would be possible to calculate the VAR for a company which holds money in a bank account? It would make sense to assess the bank financial soundness, calculating the CAP ratio e than calculating the VAR for that company? what formula might be applied for this kind of calculation? thanks!
Shouldn't the distribution be 2.5% each side with 95% confidence level instead of 5% on the left side?
He says in the video that VAR is single tailed because it is only concerned with losses
hi, shouldn't the test be two tailed?
Thank you so much
The horizontal scale: shouldn't the value at the center of the chart equal $0.0 ? Right now is $1.0.
I happened to define N(1,1) per the label σ = 1, µ = 1 rather than a standard normal, as mentioned at th-cam.com/video/mvl32w_y38I/w-d-xo.html
very nice video!! thank you!
Hi thanks for the great video, I am doing thesis using GARCH-MIDAS model using Generalized Hyperbolic distribution, do you have any idea how to compute VAR in R using these models and distribution?
Hello, completed your thesis? jedwriter.com
thank you
Hello, can i please know what the confiance level is?
We select the confidence level. Typical selections are 95.0%, 99.0% or 99.0% because we are typically interested in levels of loss that shouldn't be exceeded except rarely; e.g., 99.0% confident VaR means "we expect this loss to be exceeded only 1.0% of the time."
Thank you!!!!!!
My prof should actually be fired as I learn more with youtube than with him
Davis Maria Davis Deborah Davis Betty
waste explanation.