[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[amibroker] Re: Correctly backtesting a system



PureBytes Links

Trading Reference Links

I would suggest two things to you ...

1.  I hear what you are saying about news driven events whether they 
are geo-political or earnings related, but I think it best to at 
least know the effects these are likely to have otherwise one may be 
in for a large gotcha when trading with real dollars.

2.  Don't limit your testing to the bear market.  If it isn't already 
then sooner or later this one will be over after which time the 
system you are building & testing will have no "experience" with bull 
market conditions unless you test it during some length of time pre-
2000.

These are just suggestions ...

--- In amibroker@xxxxxxxxxxxxxxx, uenal.mutlu@xxxx wrote:
> Hi Fred,
> on the other hand such events make it IMHO 
> impossible to "successfully" backtest a system
> because they simply "disturb" the testing process,
> and having a big impact on the overall performance, IMHO. 
> 
> I have the feeling that using this method tells more
> about a system's quality in the general case.
> 
> I must admit, I'm using my own backtester, which 
> is somewhat much different by design than that of AB.
> There I simply specify the initial capital and define
> on maximally how many securities it shall invest the
> whole capital. And, the selection process is based on 
> individual scores the system calculates for each stock 
> (as usual based on some indicator values etc.) and then 
> picking the n best scoring for the actual test trades.
> 
> I tried the same code in ABs backtester (execept sorting/ranking, 
> because this is not possible in AB) also over the N100 and it gives 
> about 14% profit since 10/1/2001 whereas using the "sorting&ranking"
> method  gives about 71%. In both cases the mentioned filter was 
> applied, and also using same commissions etc.
> I'll do some more experiments.
> 
> UM
> 
> ----- Original Message ----- 
> From: "Fred" <fctonetti@xxxx>
> To: <amibroker@xxxxxxxxxxxxxxx>
> Sent: Monday, April 28, 2003 4:26 PM
> Subject: [amibroker] Re: Correctly backtesting a system
> 
> 
> > UM,
> > 
> > Personally I don't like arbitrarilly throwing away trades based 
on 
> > what your suggesting there i.e. moves of +/- 20%.  There are no 
ways 
> > I know to avoid future price spikes because of news events or 
> > whatever in real life so IMHO they best be dealt with in trading 
> > systems as well.  However to answer your question, that statement 
had 
> > no effect in most of the systems I use until I reduced the 
numbers 
> > down to +/- 4.x% at which point it started to have negative 
effects.
> > 
> > --- In amibroker@xxxxxxxxxxxxxxx, uenal.mutlu@xxxx wrote:
> > > Hi,
> > > 
> > > in the group "amibroker-ts" I've posted IMHO a very 
> > > important finding on how to correctly backtest a system.
> > > I would like to ask people if they could apply the one line
> > > AFL code on their trading systems and report if the 
> > > performance of their trading system significantly changes.
> > > 
> > > I also would ask what others think whether such a procedure
> > > during backtesting should be applied or not.
> > > 
> > > PS: there was a corrected version of the formula posted
> > > 
> > > Thx
> > > UM


------------------------ Yahoo! Groups Sponsor ---------------------~-->
Get A Free Psychic Reading! Your Online Answer To Life's Important Questions.
http://us.click.yahoo.com/O10svD/Me7FAA/AG3JAA/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/