Tracking error VAR
- Thread starter ajsa
- Start date
sudeepdoon
New Member
Hi David,
Just addiing another dimension to the question:
Does it have a link to the Tracking Error we have already read ?
Just addiing another dimension to the question:
Does it have a link to the Tracking Error we have already read ?
Hi ajsa & sudeep,
I just uploaded 7 min briefcast on tracking error (at the end, a bit on tracking error VaR)
http://www.bionicturtle.com/learn/article/tracking_error_7_min_video_foundation/
(it is based on member learning XLS @ http://www.bionicturtle.com/premium/spreadsheet/1.a.7._simulation_tracking_error_sortino/)
Amenc's tracking error (TE) = Jorion's tracking error volatility (TEV) = standard deviation of series: (portfolio returns - benchmark returns)
and this should be our (FRM) definition...
a confusion arises because Jorion defines tracking error as portfoilo - benchmark, but this is not helpful to us. For us (consistent with Grinold), the definitions are:
active return = portfolio return - benchmark return, and
tracking error = standard deviation (active return) = Jorion's TEV
and the VaR is similar to our typical VaR, we can derive two ways:
1. parametrically, if we want to impose normality = TE * NORMSINV(95%). See how is simply "scales the standard deviation by the normal deviate", or as i briefly mention in the video,
2. we can just take a PERCENTILE() of the series of active returns; so now we are non-parametrically asking "what is the worst likely active return, given some confidence"
i hope that's helpful...David
I just uploaded 7 min briefcast on tracking error (at the end, a bit on tracking error VaR)
http://www.bionicturtle.com/learn/article/tracking_error_7_min_video_foundation/
(it is based on member learning XLS @ http://www.bionicturtle.com/premium/spreadsheet/1.a.7._simulation_tracking_error_sortino/)
Amenc's tracking error (TE) = Jorion's tracking error volatility (TEV) = standard deviation of series: (portfolio returns - benchmark returns)
and this should be our (FRM) definition...
a confusion arises because Jorion defines tracking error as portfoilo - benchmark, but this is not helpful to us. For us (consistent with Grinold), the definitions are:
active return = portfolio return - benchmark return, and
tracking error = standard deviation (active return) = Jorion's TEV
and the VaR is similar to our typical VaR, we can derive two ways:
1. parametrically, if we want to impose normality = TE * NORMSINV(95%). See how is simply "scales the standard deviation by the normal deviate", or as i briefly mention in the video,
2. we can just take a PERCENTILE() of the series of active returns; so now we are non-parametrically asking "what is the worst likely active return, given some confidence"
i hope that's helpful...David
...in case it's helpful, here is the XLS i used for the video (the TE VaRs are highlit in RED):
http://sheet.zoho.com/public/btzoho/trackingerror-v1
David
http://sheet.zoho.com/public/btzoho/trackingerror-v1
David
Hi asja,
yes, in theory for benchmarking, but it's got real problems; e.g., if you porfolio is a constant +1% above benchmark, that could give TE = 0 due to lack of variability, so in practical terms, i'm more aware of its dis-uses
yes, as you have; or as i suggest above, the worst percentile of the series of {active return1, active return2, active return3 ...}
VAR(porfolio) - VAR (benchmark) does not work b/c it loses the cross-sectionality; e.g., they could be inversely correlated and you could get 0 TE with this
David
yes, in theory for benchmarking, but it's got real problems; e.g., if you porfolio is a constant +1% above benchmark, that could give TE = 0 due to lack of variability, so in practical terms, i'm more aware of its dis-uses
yes, as you have; or as i suggest above, the worst percentile of the series of {active return1, active return2, active return3 ...}
VAR(porfolio) - VAR (benchmark) does not work b/c it loses the cross-sectionality; e.g., they could be inversely correlated and you could get 0 TE with this
David
Hello - the above link doesnt' seem to work, is it somewhere else in the forum or online? Maybe more importantly, is this something we need to know for the exam? It is mentioned in the following AIM, but I'm not sure how in depth we need to get:
"Explain how VaR can be used as a performance benchmark."
Thanks,
Ryan
Fund value * Active risk * deviate
Seems pretty straight forward actually.
Seems pretty straight forward actually.
Hi
Yes Var can be used in performance benchmark. In performanve ratios we could use Var as a risk benchmark for eg excess return/Var gives return per unit of Var(risk) and compare performance of the funds based on this measure. Yes Var for the fund following a benchmark could be calculated by seeing max fund can loss for deviating from benchmark which is Var at a given CL is deviate*stddev(te)*value of fund is what u cited above. Now use Return-benchmark return/Var to compate funds following a benchmark. But for performance benchmark Var can certainly be used as a risk factor in return/risk performance benchmark for funds.
Thanks
Yes Var can be used in performance benchmark. In performanve ratios we could use Var as a risk benchmark for eg excess return/Var gives return per unit of Var(risk) and compare performance of the funds based on this measure. Yes Var for the fund following a benchmark could be calculated by seeing max fund can loss for deviating from benchmark which is Var at a given CL is deviate*stddev(te)*value of fund is what u cited above. Now use Return-benchmark return/Var to compate funds following a benchmark. But for performance benchmark Var can certainly be used as a risk factor in return/risk performance benchmark for funds.
Thanks
Ashok_Kothavle
Member
Dear Mr David,
Are these links available?
http://www.bionicturtle.com/learn/article/tracking_error_7_min_video_foundation/
http://www.bionicturtle.com/premium/spreadsheet/1.a.7._simulation_tracking_error_sortino/
http://sheet.zoho.com/public/btzoho/trackingerror-v1
Regards
Ashok
Are these links available?
http://www.bionicturtle.com/learn/article/tracking_error_7_min_video_foundation/
http://www.bionicturtle.com/premium/spreadsheet/1.a.7._simulation_tracking_error_sortino/
http://sheet.zoho.com/public/btzoho/trackingerror-v1
Regards
Ashok
@Nicole Seaman can you run interference here a bit (I assume @Ashok_Kothavle is a paid member?), thanks. I just completed the revised Amenc XLS which will shortly be uploading to the study planner; e.g., it includes an improved simulation of the Sortino
Hello @Ashok_Kothavle
The links that you have posted are very old, and I'm not sure where you got them from, but they look like they were previous materials that we offered (pre-2013?). Any of these materials that are still relevant to the curriculum are published in the study planner.
Since you are a paid member for Part 2 materials, you will find some of this content under the Andrew Ang reading under Topic 8, Risk Management & Investment Management, in the study planner. Ang, Chapter 10 covers the following learning objective: Define and calculate alpha, tracking error, the information ratio, and the Sharpe ratio.
Amenc, which is in Part 1 of the curriculum under the Foundations of Risk Management topic, covers the learning objective: Compute and interpret tracking error, the information ratio, and the Sortino ratio. Part 1 materials are only available with a paid Part 1 subscription, so you will not have access to any Part 1 materials unless you purchase a subscription.
I hope this helps you to find what you are looking for in the study planner.
Nicole
emilioalzamora1
Well-Known Member
Honestly speaking this is a very difficult topic and should be treated with highest caution.
I just came across a paper by Richard Roll called 'Mean/Variance Analysis of Tracking Error.'
https://www.google.at/url?sa=t&sour...HBBfHOrV2i9RX64DA&sig2=9B5KWsZWr2lzGCJo3E17Lw
Need some time to get through this and digest to come up with valuable conclusions.
At first glance I can tell that Carol Alexander 'Market Risk - Practical Financial Econometrics' has a good section devoted to Tracking Error (page 33 onwards). Please allow some time to tell what she wants to convey.
I just came across a paper by Richard Roll called 'Mean/Variance Analysis of Tracking Error.'
https://www.google.at/url?sa=t&sour...HBBfHOrV2i9RX64DA&sig2=9B5KWsZWr2lzGCJo3E17Lw
Need some time to get through this and digest to come up with valuable conclusions.
At first glance I can tell that Carol Alexander 'Market Risk - Practical Financial Econometrics' has a good section devoted to Tracking Error (page 33 onwards). Please allow some time to tell what she wants to convey.
Last edited:
Similar threads
