BRW Value-at-Risk - improving historical simulation (Excel) (SUB)
HTML-код
- Опубликовано: 17 апр 2020
- How to account for recency bias in historical simulation VaR? This video implements the Boudoukh, Richardson, and Whitelaw (1997) procedure, applying exponential moving average to assign higher weights to recent historical observations.
Don't forget to subscribe to NEDL and give this video a thumbs up for more videos in Finance!
Please consider supporting NEDL on Patreon: / nedleducation
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
You are awesome! Thank you. This method has so many names ( such as Weighted Historical Simulation, Hybrid Historical Simulation, and now yours)
Excellent video! Such a simple and easy-to-follow explanation! Thank you.
Very valuable.
My professor uses formulas in slides to explain this method and other VaR methods, which I read many times but still haven't grasped it. And I studied at a business university. This video explains everything in 10 minutes. Why don't universities organize practical sessions like this ?
Thank you for the excellent video. Sir, please would you explain how the BRW approach is being applied to multi assets portfolio?
Hi Murad, and happy you liked the video! The BRW approach can be applied to a portfolio in exactly the same way, simulating portfolio returns using historical data akin to the conventional historical VaR and then applying the technique covered in the video.
i have a question on ; what is the minimum of history data to estimate the VaR ?? IS IT one year or 2 according to basel committee ?
Hi Omar and thanks again for the question :) BCBS has been historically inconsistent on their advice regarding VaR backtesting, currently the requirement is 250 trading days (i.e., one trading year). Hope it helps!
can you do a video on Filtered Historical Simulation VAR?
tHANKS SIR, the best explanation ever, i have a question:
if we apply the formula of:
(1-Lambda)* lambda power n-1
we'll obtain the same results ??
Hi Omar and many thanks for the question. Yes, this formula should produce the same results as it implicitly "weights" the lambdas in the first place. Feel free to use this approach instead if you think it is more convenient for you.
@@NEDLeducation thanks sir
Excellent! Thank you. Could you talk about portfolio optimization with higher moments?
Hi Ronald! This is an interesting topic, we will definitely consider it for a video in the near future!
Hello there.
Great video thank you very much.
Would you know how to calculate this age weighted VaR with a moving window of 250 days in Excel? I already have a EWMA Moving window (parametric) that calculates moving volatilities but i am still unsure how to calculate the Age weighted HS in Excel.
Maybe you could help there :)
Hi David, and glad you liked the video! As for your question, it is much less straightforward to calculate the time-weighted HS VaR than it is to calculate the time-weighted parametric VaR as it most often requires sorting the observations from lowest to highest for every VaR calculation. However, it is possible to do it using a different spreadsheet configuration. Might do a video on rolling VaRs using different approaches in the distant future. Hope it helps!
Thanks for your presentation, it’s very useful, just one question, I think there is another semi parametric model, hw , it seems this model takes more attention on the state of market, the forecast volatility , for you which model is better ? thanks for your answers
Hi Ping, and thanks for the feedback! As for your question, not sure which alternative model are you implying, there are many tweaks and varieties of VaR, can comment in greater detail if you elaborate further. Overall, BRW VaR can be useful if you want to apply historical VaR to a very long sample (5+ years, for example), but are concerned that earlier observations can be irrelevant to the current state of the market (for example, a company growing from small and very volatile to a larger and more stable). In general, historical simulation is useful both as a sanity check for your parametric model (if it gives a drastically different result to a parametric estimation, there is something wrong with the distributional assumption), or as your main VaR model if you are unsure whether any theoretical distribution fits the data but are reasonably sure the historical distribution is stable through time. Hope it helps!
@@NEDLeducation thinks for your answer, it’s more clear for me . By the why , the hw model is hull&white var model, Incorporating volatility updating into the historical simulation method for VaR
And in the current situation, the volatility of underlying moves very high, I don’t know in the future markets regulations if they will adjust the var with this model...
What is Brw method advantage of it and what is the significance of lambda
How to interpret the results
Could you please explain asap
Hi Phani, and thanks for the question! The BRW method is named after the initials of the authors (Boudoukh, Richardson, and Whitelaw, 1997), and it is a modification of historical simulation where you put more weight on more recent observations. Lambda is the decay parameter that shows by how much the relevance of observations decrease with time.
Do we have to take the closing price or the adjusted closing of the stock?
Hi Mohteshum, and thanks for the question. You can take either, however the adjusted close takes into account dividend payments and thus is more representative, especially over longer time horizons.
@@NEDLeducation Thank You
Isn’t exponential just an arbitrary weighing scheme though?