|
PureBytes Links
Trading Reference Links
|
Ken,
Thanks for your comments. I'll try not to let you down.
To a degree, all of this "stuff" about historical lists came
about when I posted an observation about a system I was going to use to
trade N100 stocks. Here is a bit of a summary of those
events:
1. I continuously read on this board and others (such as the
Holygrailism board) about people who have backtested some new fangled
system against the current N100 list.
2. Up until a few weeks ago, I never looked at systems that only
traded a finite basket of stocks, such as the N100.
3. I was frustrated by posts on various boards where people were
talking about getting backtested results showing 100%, 200% and more in
the way of annual returns.
4. I put together what I thought would be a good system to trade
the N100 stocks and started my research with the current list since it
was easier than using the historical lists.
5. I also got performance figures in the 100% CAR range with a
reasonable charge for slippage.
6. I was about to start trading this system with real money, but I
decided to try replacing my "secret" logic with a simple,
random entry method. I do this with any system before
deciding to trade it.
7. The random method also made 100% CAR. So, my
"secret" system was no better than random entry.
8. Then, I put the effort into using forward-looking historical
lists rather than a backward-looking current list.
9. CAR went from 100% to 20% and my bubble was burst.
10. I then started a thread on the subject and started sharing the
historical lists.
You may wonder why I share the lists? Mostly because of the
amount of work that goes into getting them accurate. It takes
hours to complete each list and it would be a shame to simply file them
away on my PC. So, at the risk of people telling me that the
work isn't worth the effort, I share them.
As to whether you should bother coding the additional logic into each
system, I have to say emphatically "yes".
We see examples of what I think is the wrong thing to do all the
time. People say things like "I sit down every three
months and prepare a watch list of stocks that have done well over the
last year"... or something equally concerning to me. They
might be looking for high yield stocks, low P/E stocks or simply stocks
that have gone up the most. What a biased list?
Why not prepare such a list based on what you knew a year ago and then
test your system on that list going forward one year with real
data? That will give you a much better idea of how well
your system works than taking a list of good performing stocks and then
backtesting to see how well your system did with that list.
Surely the later must seem silly in the context I am describing?
At 11:15 AM 1/20/2004 -0500, you wrote:
Chuck:
thanks for the thoroughness of your reply your consistency, diligence,
followup and general overall helpfulness has no equal on this list or any
other I have come across&.many thanks.
This subject is proving to me to be one of the more complex to visualize
an end result or consequence. I am gradually moving from general
disbelief to at least planning on devoting time for testing to see for
myself . Trouble is, I right now am fuzzy about what I will see ,
and thus confused about the learnings and benefits that will
accrue. Thus I asked the question which you answered and the answer
was unexpected.
&worse results more often than not&. , but more accurate
results??? Thus, a system tests to an x result on current data but
to a y result on past segregated data, so how do you decide to move
forward?? Set minimum response AA from the segregated results and
then if this is exceeded, you can have high expectations of a good result
moving forward with real trades?? Boy, it is very circular for me,
and I suspect only a long series of tests then observations will give me
a feel for the value of this kind of construction.
If anyone has a clear and concise view of how this is a requirement ,
much less just a nice to have , then I would sure like to know.
What about all the folks (surely there are some), who blindly backtest
using current data, find a set of equations that give good results (using
historical rules of thumb), trade those results moving forward, and make
lots of money (or some money----or satisfactory money for
them)?? Could these folks do better if they knew and
adopted this segregated list approach? Or not?
Grrrrrrr.
But, Chuck, do not let my confusion and frustration confuse the fact that
I appreciate, as I said above, all that you do for us here.
Ken
-----Original Message-----
From: Chuck Rademacher
[mailto:chuck@xxxxxxxx]
Sent: Tuesday, January 20, 2004 10:14 AM
To: amibroker@xxxxxxxxxxxxxxx
Subject: RE: [amibroker] Russell 2000 constituent lists
Ken,
I failed to answer your last question which was
"Running such a segregated
list approach vs just the current list should show much improved results,
am I correct?".
IMO, backtesting this way would give you
worse results more often than not, at least with a long-only
system. Think about it. Compare the results of a
"buy and hold" approach based on the 2000 list compared to the
2003 list, starting in (say) 2000. The 2003 list contains the
best performing stocks. By definition, the results using a
historical list should be worse.
The important thing, to me, is that the results should be more
accurate. In spite of saying that I would expect worse
results using historical lists "more often than not", many
systems will perform better simply because you would be including the
de-listed stocks and, therefore, providing a larger universe for your
system to consider.
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
Yahoo! Groups Links
To visit your group on the web, go to:
http://groups.yahoo.com/group/amibroker/
To unsubscribe from this group, send an email to:
amibroker-unsubscribe@xxxxxxxxxxxxxxx
Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service.
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
Yahoo! Groups Links
To visit your group on the web, go to:
http://groups.yahoo.com/group/amibroker/
To unsubscribe from this group, send an email to:
amibroker-unsubscribe@xxxxxxxxxxxxxxx
Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service.
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
Yahoo! Groups Sponsor
ADVERTISEMENT
Yahoo! Groups Links
To visit your group on the web, go to:http://groups.yahoo.com/group/amibroker/
To unsubscribe from this group, send an email to:amibroker-unsubscribe@xxxxxxxxxxxxxxx
Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service.
|