|
PureBytes Links
Trading Reference Links
|
My previous post on this topic was actually a followup to an earlier
post that talked about managing risk when position sizing could not
be used. Since some might be interested in this, a copy is given
below.
If anyone has a different method for managing risk when using
baskets of stocks with market timing signals, please share. I would
be very interested.
Oh, there is a bit of off topic stuff (dealing with VV-VectorVest)
in the following. But most deals with risk management in a generic
way.
b
------copy------
From: "b519b" <b519b@xxxx
Date: Sat Aug 9, 2003 12:15 pm
Subject: How to Test VV strategies (for John & other newcomers)
http://groups.yahoo.com/group/vectorvestonlineusersgroup/message/2927
5
John,
Welcome. You will find this forum to be very stimulating and
educational. I do. Many disagree with me at points. I often learn
the most from trying to understand why someone would take a
different perspective than the one I start with.
Disclaimer: The following is not intended as trading advise. It
should be considered as being "educational" and only a partial
education! You are responsible for your own trading decisions. You
are also responsible to ensure that your education is sufficient for
the type trading you plan to do.
My comments will emphasize risk management because this topic is
very profiable yet seems to get less attention than I think it
deserves.
I agree with Dennis [Dennis Fluegel] that the place to start is with
a simple timing system (MTS) and then use these to test some stock
selection strategies (SSS). Dennis likes a MTS based on the MTI/VVC
combination. My preference is to base my MTS on something like the
RUT or VLIC which will continue to work if VV changes the MTI (as it
did a couple of years ago). I can give you trade dates for a MTS
based on the RUT if you are interested.
Whatever MTS you decide to use, here are a few things to keep in
mind.
1. There are many "good" MTS but no perfect one. Expect "false"
signals. There are many workable MTS approaches. Dennis has one. I
have another. We can both be happy.
2. Expect "whipsaw" losses. There usually are small losses (and
sometimes large ones) that come from the false signals. When you
back test various SSS, be sure to record the losses of your
portfolio. Everyone records the average gain. This could be called
AAR (Average Annual Return). Those who are smart also record the
losses. These losses can be analyzed to determine a SSS's MDD
(maximum draw down).
3. Plan to be a "survivor". To be a winner, you have to be a
survivor. In practical terms, you have to be able to "survive" the
worst series of losses that your MTS and SSS combination can throw
at you. You also need to know your MRT (maximum risk tolerance).
Each person's MRT will be different. For some, their MRT will be
10%, for others it may be 25% or 33%. You can expect your MRT to
change as your financial circumstances change. It is vitally
important to know what you current MRT is. If you do not, your
emotional "gut" may kick in a the worst possible time.
4. Once you know your MRT and the MDD of your MTS-SSS system, you
can determine what percentage of your trading capital to use.
An illustration might help at this point.
Let's say a person determines his MRT is 40% in financial terms.
That means he/she could financially survive a 40% loss in trading
capital (trading capital is not necessarily total capital). However,
the same person knows that a loss of 20% would likely cause him/her
to stop trading. Thus the person's MRT is really the lower of the
financial MRT and emotional MRT. In this illustration, the MRT would
be 20%.
Now for the MDD. Let's say after sufficient testing (again what
is "sufficient" will vary from person to person), you have found a
SSS that has a great AAR of 120%/year and its MDD is 40%.
What percentage of the person's trading capital should be used with
this SSS? The math is simple: MRT/MDD or 20%/40% = 1/2. So it would
be a "reasonable" risk (still a risk - one can never totally
eliminate risk) for a person with a 20% risk tolerance to use one
half of his/her trading capital with the SSS. Of course, this will
reduce the expected gain from 120% to 60%. Here is the math:
AAR * MRT / MDD = a reasonable expectation.
120 * 20 / 40 = 60
Does this make sense? If so, consider the following.
How much could such a person make trading a SSS with an AAR of 90%
and a MDD of 15%? The math may surprise you.
AAR * MRT / MDD = a reasonable expectation.
90 * 20 / 15 = 120.
How does one get a 90% system to give 120%? By using a bit of
margin. When is it reasonable to use margin? When one's MRT is
greater than the MDD. Of course, no one has to use margin. And no
one should ever use margin without having an estimate of the
system's MDD -- and being very confident that that MDD estimate is
accurate!
As you record the results of your tests, you may wish to give each
SSS a DDAR (draw down adjusted return) or MAR (market adjusted
return) number. Both are simply AAR / MDD. I like the concept of the
MAR which I learned for a non-VV discussion board. My mind finds
DDAR to be a bit easier to remember, but the concept is the same.
Now there is a lot more to consider that just the above. But the
above will get your education started.
What else is there to consider? Here are some that quickly come to
mind. How does the number of stocks in a portfolio affect AAR and
MDD? Generally using fewer stocks will increase the AAR but also
increase the MDD. Using more stocks could be more profitable when
MDD and MRT are considered.
Also, how good are the estimates of ARR and MDD? Testing can only
give estimates and estimates have margins of error. So, how much of
a margin of error should one allow for in the AAR, and MDD given the
number of market timing signals studied in a test. Generally, the
greater the number of distinct tests, the better the estimate. Also,
one may wish to adjust estimates due to "survivorship" biases
and "stock split" biases that may be a data base. This will affect
one's confidence in the MDD and AAR estimates. In addition, has the
MDD taken into account the possibility of a series of losses back to
back. It is the series of losses that matters, not just a single
loss. Could the future hold a longer series of losses than past
testing has revealed? How much should the MDD estimate be increase
to allow for a rougher ride in the future?
Has the SSS been "curve-fit"? Has the MTS been "curve-fit"? Are the
SSS and MTS "robust"? Do they still work if their key parameters are
altered slightly? How much does the MDD increase as the key
parameters are changed? How much should the MDD estimate be
increased?
b
-----------------
------------------------ Yahoo! Groups Sponsor ---------------------~-->
Special Sale: 50% off ReplayTV
Easily record your favorite shows!
CNet Ranked #1 over Tivo!
http://us.click.yahoo.com/WUMW7B/85qGAA/ySSFAA/GHeqlB/TM
---------------------------------------------------------------------~->
Send BUG REPORTS to bugs@xxxxxxxxxxxxx
Send SUGGESTIONS to suggest@xxxxxxxxxxxxx
-----------------------------------------
Post AmiQuote-related messages ONLY to: amiquote@xxxxxxxxxxxxxxx
(Web page: http://groups.yahoo.com/group/amiquote/messages/)
--------------------------------------------
Check group FAQ at: http://groups.yahoo.com/group/amibroker/files/groupfaq.html
Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
|