Saturday, July 15, 2006

Predator is born!

Hi all, thought i'd just make some comments - been a week or so since ive posted so i thought i'd let you know what i've been up to.

Over the past three months ive been refining and testing a new trading system. - it's purely mechanical which is suprising because it does use indicators and i always had the belief that it was virtually impossible to build a mechanical system which used indicators. even more so I didnt believe in automatic trading.

Anyway, to cut a long story short I fell onto this system by accident. Ive slowly been testing an indicator for a while now based on a single setting - it was looking good to produce prob 80-100 pips a week which is good in my eyes.

Just on a side note to tell you that with mechanical systems my nirvana is around 100 pips a week with a low drawdown. I thought if i could make that then it would be a great system to sit aside Grail.

Well, that nirvana i think was a limiting belief - i didnt think it was possible to build a mechanical system that did much more than that and i guess that may have held me back in the past.

That belief has now been not only broken, but smashed into pieces.

During my testing i thought it would be much easier in the interim to have an mt4 expert to backtest the system so i could do many calculations before hand testing. so, thanks to Fluty, Trevor and Myself the 'Predator 'expert was born.

After a few tests which looked ok i wondered just how many different parameters i could get the backtester to test - so into the code i went and placed code so that i could do just that.

when i ran this test the number of combinations was around 117,000. I have a fast pc but even then it was to take over 7 days to do the test.

Well, when the test completed i couldnt believe the results, they were amazing.

Now i know what you are thinking, ive seen many backtests which look great and turn to crap when you run them - ive seen them too.

I can remember an expert called AIP (alwaysinplay) that did millions in the backtest but when you ran it live crashed your account quicker than an israeli retalliation.

So, when i'd optimised this and put it on seven pairs on a demo account i believed it would crash and burn.

It didnt,

well, not yet anyway.

Ive only ran it for a week so this isnt a good enough sample size but get these results

account started at $10,000

186 trades taken

155 wins

31 losses

win percentage 81.87%

Overall Pips Gained 534

Cash gained $6133.41

Gain on account 61.3%


Now ths looks pretty damn good i'm sure you will agree but i guess time will tell. Nevertheless I am swiftly opening an MT4 account to trade this with some real money and i'll keep you updated with the results from the test until the live account is open.

ps, before anyone thinks this is some slick marketing - it's not for sale nor will it ever be.

keep tuned in :)

4 comments:

spahiu said...

sounds almost to good to be true... i think u already know that backtesting through MT4, even with imported data, has many flaws... so forward testing is the way to go...
since u "felt over this system by accident" how about pointing in that direction :P ..ofcourse if the inspiration source is available on some public forum ...

Soultrader said...

no, what happened was we were designing an indicator for the four hour charts which also looked at the daily charts - whilst doing this we accidentally ran a test on the 5 min chart with outstanding results - as the work on the two indicators involved was done by my team i'm afraid i cannot share apart from it's a 'happy accident'

spahiu said...

no worries,

Luke Sullivan said...

Hi soultrader,

Interesting results. I have been reading your forum for sometime now (from aus) and I must say its been a pleasure. I love trading and have little experience with forex. Having just moved to London from Australia I could not help but think of you to contact regarding ofrex trading. Do you offer coaching at all? I am a programmer by trade (Java) and are currently trying to develop a suite of MT4 experts. Being a programmer I obviously think automation is the way to go.

If possible i would love to have a chat with you offline.
sullivanluke@hotmail.com

Thanks,

Luke