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

[amibroker] Re: Correctly backtesting a system



PureBytes Links

Trading Reference Links

This brings up an interesting point or two ...

1.  Unless you are testing with fixed position sizes the outliers 
should be judged based on % changes as opposed to dollar changes.  In 
fact % changes will work equally well for both.

2.  I will agree that it would interesting to see if one removed all 
the outlier trades what the results were like without reoptimizing 
and then again with reoptimizing and then after reoptimizing looking 
at the system performance again with all the trades back in.

Thought ?

--- In amibroker@xxxxxxxxxxxxxxx, "Anthony Faragasso" <ajf1111@xxxx> 
wrote:
> Um,
>  
> Reading your recent post...Brings to mind what I call Outlier 
Trades.
>  
> Outliers have some unusual property, such as excessive profit or 
loss.  The
> amount
> of excess is measured by the number of standard deviations away 
from the
> typical ( average )
> value.  For example, if a system's average profit is $ 200 per 
trade and the
> standard deviation
> is $ 50, then a trade that produces $400 profit would be ( (400 - 
200 ) / 50
> ) or 4 standard 
> deviations from average.
>  
> Outliers are usually considered to be 3 or more standard deviations 
from
> average.
> Essentially, they are aberrations that cause system results to be 
unfairly
> positively or negatively
> biased.  By removing outlier trades from the evaluation process, a 
new net
> profit figure can be
> calculated.  This new select net profit  figure, devoid of all 
aberrations,
> may offer a cleaner, more 
> realistic trading perspective.  Obviously, Trend-following systems 
designed
> to accept numerous small
> losses to capture the infrequent big move rely on Outliers to be 
profitable.
>  Therefore, whether or not
> it is better to remove outliers requires some thought.
>  
> In general, systems that are heavily dependent on outlier trades 
have
> artificially inflated or deflated
> net profit results.  Since Outlier trades generally do not reoccur 
on a
> regular basis, they should be
> removed to present a more realistic trading perspective.  The goal 
is to
> find a system with a select
> ( nonoutlier ) net profit figure worthy of trading.
>  
> Example:
>  
> Is a system worth trading if it only generates $ 400,000 in profit 
without
> outlier trades as compared 
> with its original $ 500,000 ?
>  
> Anthony


------------------------ Yahoo! Groups Sponsor ---------------------~-->
Get A Free Psychic Reading!
Your Online Answer To Life's Important Questions.
http://us.click.yahoo.com/cjB9SD/od7FAA/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/