Great EA in backtest! - page 68
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
hi David
What is file set for this version?
tnx
regardsnot sure what file set means. It is 85f with my settings and the trailing s/l fixed.
If thats what you mean.
ok. here is my best results for usd/jpy agressive.
Take the version I posted above and set ValuesPeriodCount=7; ValuesPeriodCountMax=7;
leave the rest as default. I tried every # from 5-23 and 7 was the best ballance.
I am working with different s/l now to see if 11 is optimal.
then next I'll test on euro/$
then I'll run it through 2 years on each pair.
any one care to run these settings on gbp/$ and $/chf?
Dave
Heres 2 years 90% quality. Maintained 63% wins same as the 1 year back test.
Dave
It could be done that a data file with all the news dates and times of interest is loaded into the expert on startup. This should also work for backtest. I can do this, but it could take a couple of days doing it in my spare time. I'll start on it.
That is awesome! I look forward to seeing it.
all months are very good besides first one or two, no matter what date test started
hang on....
perhaps the challenge is in interpreting the results you are getting...
what you describe I have also seen in my test results...
namely that the first one or two months the EA runs on are rough almost no matter what months you start the test on. I have a theory about this.
This EA seems to be running it's own internal simulation from which it derives statistical probabilities which influence it's decisions. It may well be that this EA has to run for a period of time to build up that statistical information before it's decisions become well grounded in that data. If that is the case (and I'm not certain yet because I've not gotten that deep into the code yet) then it's not going to give identical trades over any specific time range depending on how far in advance of encountering that time range the program started. Does that make sense?
In other words my theory is that the EA has to learn by running before it really starts winning because it has to build up it's internal statistical base.
To test this I just ran two tests one starting August 1, 2006 to today and the other starting September 1, 2006 to today. Then I put the results side by side and noticed that there are slight differences in the trades entered in Sept. and Oct. It's possible that this is evidence of the theory being right. There may be other causes but this could support the theory.
I have a theory about this.
i thought about it too, but look at kat's or nikkeifx's backtests and try to compare them to one of mine. with the same parameters i'm getting very good results and both, kat and nikkeifx, don't.
if CT collects statistic in variables (i'll try to check it out) it shouldn't be a problem to run a backtest, check the statistic, and put them as parameters to initialize CT for another test.
Argorn,
what is it you found the stop loss index does when changed? higher index verses lower index?
I have the conservitive version on euro$, and agressive version reciently tested on $jpy running on $jpy. Both running now on real money.
I also opened a demo and am running the agressive version on 8 different pairs for forward testing.
I am getting really pissed at how I can get good results on these 2 pairs and when I do anything on any other pairs it sux.
The news times are screwing up my tests from what I can see on the charts visual.
So we'll see what forward tests bring on the 8 pairs. with the same 11 pip s/l I use on every 1 hr chart.
Argorn, what is it you found the stop loss index does when changed? higher index verses lower index?
I'm not certain because what it did on some tests was reduce the relative draw down. It has not however proved to do so on ALL tests since I have changed it. So it's hard to tell.
This is what I think I'll run in live forward this week at least to start with...
I have these two reports one with maxlots=10 and the other with maxlots=0.5
I have the settings I'm using in this EA attached and I'm using it on the one hour gbpusd chart on Interbank mini account.
As you know my account is only got $368 in it. I have a hard time letting it trade with lots any more than 0.5. On this level trades will be between $4.50 and $10.50 one way or the other and I suspect between two and four trades a day on average.
I can see that I may have to tolerate a drawdown of $60-$80 dollars which is what the drawdown was from sept 27 to present which represents one of the worst time ranges of data to date. I think I can stomach that if I have to.
I just don't have the guts to pull the trigger and allow it to trade with maxlots=10 which I can see could potentially really grow the account. I don't know perhaps I should build a stepping control that would allow the lots to increase what I think I could stomach based on the freemargin growth. Actually I just need to find a risk setting still that I can live with and that would do it. I must say I find that 3.73% relative drawdown on the 0.5 maxlots comforting while I find the other one more exciting for growth....
decisions decisions...money management questions...hum