|
PureBytes Links
Trading Reference Links
|
<From my porch I don't trade systems that don't have 200+ trades in
them end to end.>
Yeah...
I always start with a full 10 or 11 yr backtest that generates
thousands of trades AND which is profitable.
Then, since I usually can't afford to do all the trades generated, I
start to 'worry the system to death' by gently increasing the
filtering criteria, being careful not to over 'curve fit' with my filters.
And finally, I try to derive a 'filtered' trading system that picks
the best representative trades using logical criteria, including trend
and relative strength measurements that isolate the best and strongest
prospects from the universe of trades.
Some would call it 'curve fitting', but I belive that if you can
rationalize the predictive ability of your filters then 'curve
fitting' is minimized.
Phsst
--- In amibroker@xxxxxxxxxxxxxxx, "Chuck Rademacher"
<chuck_rademacher@x> wrote:
> Thanks, Graham. I concur with everything you've said.
>
> For what it is worth, however, I wouldn't trade a system with three
or four
> variables based on less than 3,000 trades... howver many days, weeks or
> years of data that requires.
> -----Original Message-----
> From: Fred [mailto:fctonetti@x...]
> Sent: Saturday, March 29, 2003 10:19 PM
> To: amibroker@xxxxxxxxxxxxxxx
> Subject: [amibroker] Re: TESTING THE UNIVERSE ?
>
>
> Grahm,
>
> Questions and/or comments regarding your post.
>
> If you are only trading long then to me it's still imperative to test
> over bear markets as well unless you have some sort of automated
> filter that keeps you out of bear markets in which case you are still
> testing over bear markets, aren't you ? It's wonderful to look back
> in history and say ... well I wouldn't have traded then because it
> was a bear market ... but the question is when did you recognize that
> it was ? after day one ? month one ? year one ?
>
> The same applies to delisted stocks, doesn't it. No way to know much
> in advance that they were going to be delisted.
>
> With regards to holding period .vs. testing period IMHO neither 10
> years for those planning to hold 12 months or 1 year for those
> planning on holding 5 days is sufficient although the latter appears
> to be more viable as there are ~50 potential trades there and only 10
> in the first scenario. From my porch I don't trade systems that
> don't have 200+ trades in them end to end.
>
>
>
> --- In amibroker@xxxxxxxxxxxxxxx, "Graham" <gkavanagh@xxxx> wrote:
> > I don't see the need for testing multitudes of stocks that no
> longer exist
> > nor for testing over large periods of time, nor having a large
> basket of
> > stocks available on your search list.
> >
> > As far as I see all you need is a timeframe to cover various market
> trends
> > (bull/bear/sideways) and enough stocks to make it viable. I would
> also look
> > at how you trade, long only, or long and short trades, etc. If you
> only
> > trade long, why test over a bear period. Logic would tell you that
> your
> > system will not be as profitable, and thus you should maybe stand
> aside till
> > your system matches the market. Your keeping of records against a
> certain
> > criteria of win/loss ratio, or profit/loss ratio would tell you
> when your
> > system is out of synch with the market. Either you have multiple
> systems or
> > keep adjusting the one system to suit current conditions would be
> options.
> >
> > The stocks you select for the testing should be consistently traded
> > throughout your test period, so I ignore any that closed or started
> during
> > the period. Your timeframe of testing should also be in line with
> your
> > trading timeframe. The test period should be longer than your trade
> > timeframe. I can only make a stab at this as I have only tested to
> suit my
> > style, but if you intend to hold for 12 month periods, then you
> test over a
> > 10 year frame, or for holds of 5 days, I would think that a test
> over 12
> > months would be sufficient.
> >
> > The size of your stock basket for systems should be related to your
> trading.
> > A small short-term trader can only manage a relatively small number
> of
> > trades at one time due to size of capital, and the human ability to
> manage
> > them. A larger trader/investor would have a different aspect being
> able to
> > manage a larger basket due to more capital, and a longer term
> perspective
> > easier to manage more trades. So if you can only manage 2 trades
> per week,
> > why have a system that provides 20 entry signals weekly. A trader
> who makes
> > 20 trades per week would need a system to signal more than 2, but
> would not
> > need 100 signals each week. So you would need to combine the number
> of
> > stocks to search and the trade signal system to match your
> requirements.
> >
> > At the end of all this nonsense I have said, there is one thing
> that really
> > counts. The methods and systems you use for trading must be
> suitable for you
> > personally. What one person does will probably not suit another.
> >
> > Cheers,
> > Graham
> >
> > <http://groups.msn.com/ASXShareTrading>
> > http://groups.msn.com/ASXShareTrading
> >
> > <http://groups.msn.com/FMSAustralia>
> http://groups.msn.com/FMSAustralia
> >
> > -----Original Message-----
> > From: Jayson [mailto:jcasavant@x...]
> > Sent: Sunday, 30 March 2003 7:30 AM
> > To: amibroker@xxxxxxxxxxxxxxx
> > Subject: RE: [amibroker] TESTING THE UNIVERSE ?
> >
> >
> >
> > Gosub,
> >
> >
> >
> > there will certainly be many on the other side of this discussion
> but FWIW I
> > try to define a universe that will trade best with a given
> strategy. For
> > instance my universe has certain price, average volume and market
> cap
> > requirements. I agree that some stocks have certain personalities
> that tend
> > to work best with certain systems. Others will argue that a Robust
> system
> > should work equally well in any market. The challenge with the first
> > approach is that depending on how far back you you are testing the
> > personality may be very different now than it was at the start of
> your
> > testing period, especially if you test back 10+ years. Look at
> MSFT, AOL and
> > CSCO as examples....
> >
> >
> >
> > Jayson
> >
> > -----Original Message-----
> > From: gosub283 [mailto:gosub283@x...]
> > Sent: Saturday, March 29, 2003 5:28 PM
> > To: amibroker@xxxxxxxxxxxxxxx
> > Subject: [amibroker] TESTING THE UNIVERSE ?
> >
> > Hi everyone,
> >
> > Please bear with me on this subject because
> > it's one which I have not yet found the answer
> > and one which I am not an expert. This question is based
> > on my current assumptions and is open to comment,
> > correction, or debate.
> >
> > (This has been discussed before but, as an onlooker,
> > I did not see a solution.)
> >
> > Here it is:
> >
> > What is the point of testing the whole universe
> > of stocks with a trading system if it is generally
> > understood that..
> > A) Some stocks are just not "system" tradeable
> > B) Some systems are best suited to certain markets.
> > C) Some stocks have unique "personalities" which work
> > with some trading techniques but not others.
> >
> > It seems to me that a test of the whole universe will give
> > a squewed result because the performance of the system
> > will be lowered by the "untradeables" and the ones with
> > the "wrong personality".
> >
> > I have written filters which divide up the universe into two
> > personality groups.(Good ones on the left...bad ones on the right)
> > This has helped to narrow down the basket a little.
> > But maybe there's another reason to test the whole universe
> > that I m not aware of. Any comments on this ? (for or against)
> >
> > PS: I think the focus should be on devising ways to define
> > and catagorize "personalities", then go exploit them.
> > (Definately easier said than done) ;-(
> >
> > Cheers,
> > Gosub283
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> > Send BUG REPORTS to bugs@xxxx
> > Send SUGGESTIONS to suggest@xxxx
> > -----------------------------------------
> > 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
> > <http://docs.yahoo.com/info/terms/> .
> >
> >
> >
> >
> >
> > Yahoo! Groups Sponsor
> >
> >
> >
> > ADVERTISEMENT
> >
> >
> <http://rd.yahoo.com/M=243066.2784921.4151384.1769302/D=egroupweb/S=17
> 056321
> > 98:HM/A=1377500/R=0/*http:/www.verisign.com/cgi-bin/go.cgi?
> a=b31540113206004
> > 000>
> >
> >
> >
> > <http://us.adserver.yahoo.com/l?
> M=243066.2784921.4151384.1769302/D=egroupmai
> > l/S=:HM/A=1377500/rand=329115638>
> >
> >
> > Send BUG REPORTS to bugs@xxxx
> > Send SUGGESTIONS to suggest@xxxx
> > -----------------------------------------
> > 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
> > <http://docs.yahoo.com/info/terms/> .
>
>
> Yahoo! Groups Sponsor
> ADVERTISEMENT
>
>
>
>
> Send BUG REPORTS to bugs@xxxx
> Send SUGGESTIONS to suggest@xxxx
> -----------------------------------------
> 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.
------------------------ Yahoo! Groups Sponsor ---------------------~-->
Make Money Online Auctions! Make $500.00 or We Will Give You Thirty Dollars for Trying!
http://us.click.yahoo.com/yMx78A/fNtFAA/i5gGAA/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/
|