It seems a long time since comments were made. How are we all doing? Live trading? Demo Accounting? Playing ATM poker and not speaking?
A few posts ago Anton noted back-tests work but live/demo trades don't. so far I agree. I have back-tested many scenarios but when I take my best one to demo account, I loose - so far 20% down.
To gain experience of the ATM I have stuck to one pair. Many times I see the trade heading to the S/L and want to intervene but let it ride to check my interpretation.
I have also tried the EMA trend filters but so far they give me a poorer result.
I find it hard (impossible) to get a set of variables to provide 5/1 consec wins along with a profit factor above 1.5 (1.57) and with low draw-downs. My DD's rise up to 13%. (but the NP is pretty good (huge), if only it wasn't a back-test!) )
Anyone willing to discuss further??
I have set all "example" ea settings loose on a small live account and as of a few days ago it was up 20%. Then I started messing with it and have managed to cut my wins in half.
ReplyDeleteWhile that was trading in the background, I have tried backtests on a few pairs and am finding the same problem that you are, namely excellent returns but high drawdowns. This appears to be repeated regardless of the time-frame I choose so I must conclude that the pairs I have chosen require a different technique. I have been concentrating on the EURAUD because I liked the way it seemed to move on a chart...So much for that idea!
It looks like I needed more perseverance, so with that in mind I have back-tested about 100 variations using the 6 month period to 22 Feb. I then ran the best on the last two months to 3/4/12 and that gave me different results. All as was said, that what was good for last year wont necessarily be good this year. So back-tested and optimised twice and have settled on a set that now encompasses the EMA's and am currently optimising those.
ReplyDeleteI now have a PF of 2.69 : DD 5.11% : EE 0.05 : P 5 : MA2 :FRSR 1.5: TE 0.05. $10k in 2 months = $6497 profit. simple extrapolation to 12 months =$2,132,947!
So I expect one will need to keep back-testing and optimising each month to fine tune for the next month.
now that would be an idea. Get your best set for say Dec. then opti it for the known Jan data and so on up to April noting the trend changes. this might assist in pointing the way ahead.
the 2.132m figure above should read $180,000
ReplyDeleteI think you are onto something there. Maybe that is the reason that we will be receiving constantly updated data. I think constant tweaking will be required but not to enormous degrees. Each month or two, the modifications needed will likely be small ( but necessary) to maximise returns. Also, moving to the new IC Markets server will help also as the spreads there are much tighter.
Delete