Curve Fitting

[maninjapan]

Im testing different filters on a strategy and I am currently looking at daily bias. My question is, Is not trading on a specific poorly performing day of the week curve fitting? I was thinking that as long as I have a logical explanation and performance is generally poorer rather than due to a couple of massive losses, then I should be ok..

What do you think?

 

[Hook N. Sinker]

Any trading rule is curve fitting.

You might split your data up into two or three parts and backtest each part separately. Paper trade your system in real time and observe the results.

I do not know of any trading system that comes with a guarantee.

 

[maninjapan]

right, thats what Im saying, its consistantly gives poorer results. I WISH I had a system with a guarantee

 

[DT-waw]

Quote from Hook N. Sinker:

Any trading rule is curve fitting.

what a shock! someone, finally, after 10 years of existance of the 'elite' trader forums said the truth.

 

[Gyles]

I had done a study on robustness using TradersStudio on a "Triple Moving Average" system.

Please have a look at the same in the following thread:

http://www.elitetrader.com/vb/showthread.php?s=&postid=2130065#post2130065

 

[good one]

Your best bet would be to try to set up some tests to prove your logical explanation false.

 

[TraderZones]

Quote from Hook N. Sinker:

Any trading rule is curve fitting.

You might split your data up into two or three parts and backtest each part separately. Paper trade your system in real time and observe the results.

And no matter how you split up the data, it still does not mean it works in the future.