|
PureBytes Links
Trading Reference Links
|
Dimitris: thanks once again for an
innovative system concept.
The purpose of my note is to comment on it
performance. Some have said that it is better to optimize over a longer
period than just from <span
>1/1/2000<font
size=2 color=navy face=Arial> and others have said to optimize just from 1/1/2000<span
> onward. I
subscribe to the former approach and my standard optimize period is 8/1/1998<span
> to 0/21/2001. I then
test from <span
>8/1/1998<font
size=2 color=navy face=Arial> to today and from <font
size=2 color=navy face=Arial>9/22/2001<span
> to today. I use
the Tonetti enhanced equity curve, which gives some good statistics but also
shows the log equity curve. A system that is going to fail will show a clear
drop off in the OOS period.
This code gives a fairly smooth equity
curve which does not “blow up” in the OOS period. It seems to
perform better on DJ30 or even SP500 stocks but not so good on NDX stocks. Drawdowns
are still double digits. I also optimized from the period 1/1/2000<span
> forward, using the same
OOS period. Performance seemed worse in that annual percentage gains were
higher but Drawdowns were very high---30% the lowest.
What is not clear to me in your message or
code is what stock or index you place in the Buy position while you are doing
the optimization. What array is being bought and sold during the
optimization. This clearly makes a difference to the result and I am
curious as to what you used.
Nonetheless, it is an innovative new
approach and I for one thank you for sharing.
Ken
-----Original Message-----
From: Dimitris Tsokakis
[mailto:TSOKAKIS@xxxxxxxxx]
Sent: Wednesday,
February 19, 2003<span
> <font
size=2 face=Tahoma>5:24 AM<font
size=2 face=Tahoma>
To: amibroker@xxxxxxxxxxxxxxx
Subject: [amibroker] The
transcendental use of Data: An application
A.
I trade BEAS<font
size=2>.
Its performance with the smoothed D-ratio5 is poor. 3/21 winning
combinations when D1=<span
>Optimize<font
size=2>("D1",42,30,50,1); and the best
gives
some +115% in 3 years. Nothing important for 57 dangerous trades.
But, BEAS<font
size=2> may give great results, much better than
+1000% with a lot of systems.
B.
I do not trade TMPW and I know almost nothing about it.
For me, TMPW is a 5-D vector [O, H, L, C, V] in the wild StockMarket
vector space.
TMPW is in the top10 list for D-ratio5 trading over the whole N100
market, according to my earlier post.
Its D-ratio5 curve gives "good" signals for the majority of
the stocks.
For D1=42, for example, it sends <span
>BEAS<span
> profits to >+1000% and it is not coincidental,
since the whole market profits are +450% with
>75% profitable stocks.[It is not a holy grail, it is a real
fighter, you should take all the risk to trade it, but 75% is a respectable
percentage]
C.
I "borrowed" TMPW clear Buy signal on Valentines day,
Feb14 to Buy BEAS [and CSCO] on the next bar Open.
The Open was cool, the day was great, some cables traffic 15 min before
the end [life is not as easy as backtesting] plus my broker
advise to sell tomorrow, but the sell order was an order, it was a
clear take-the-money-and-run session, the system will make its own profits,
statistics is an additive science [linear or not].
CSCO will pay the expenses, <span
>BEAS<span
> will pay the research, I should learn something for
TMPW, perhaps it is a great company.
D.
I will refill my hot coffee and try to understand what I was doing
yesterday.
Dimitris Tsokakis
the Yahoo! Terms of Service.
Yahoo! Groups Sponsor
ADVERTISEMENT
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 the Yahoo! Terms of Service.
|