stevehopwoodforex.com
https://www.stevehopwoodforex.com/phpBB3/
Print view

Fratelli EA by Dragos and magft
https://www.stevehopwoodforex.com/phpBB3/viewtopic.php?t=77		 Page 16 of 23
Author:  MickyMouse [ Fri Dec 02, 2011 9:25 pm ]
Post subject:  Re: Fratelli EA by Dragos and magft
magft wrote:
MickyMouse wrote:Empty4 is good enough if you can find enough tricks to overcome stupid limitations... tradestation is better..but would`nt really vouch for it either... i think dukas is the best place... and off course the payed platforms :D

yes from 0 to 10, u force the indy to look on youre offline chart
last high on 10 mins is iHigh(NULL,10,1) for example
and so on :D
So in theory as long as your M10 history file is big enough you and run in ST on any chart if you hard code M10 in indicators?

I read you can use a trading simulator with the period converter indy running to generate the data and run the EA on the fly by configuring a template that you can drop on with all the indicators you need.

I'll give it a try this weekend. The EA is working to a point, still sometimes trades not the same as indicator this is mostly due to slippage, well that is what happend today. I lost 74pips yesterday but made 86 today, so overall i'm up 268 pips this week. Should have been more but i have been testing GU and it lost 60 pips this week. I have found 8-9 and 13-14 the best times to trade so far for EU,AU and EJ as no losses in this time slot.
yes if u hardcode the indy to look at 10 mins, it will look there...
don't know about the trick with period converter, but im interested at what will show
Author:  magft [ Fri Dec 02, 2011 9:34 pm ]
Post subject:  Re: Fratelli EA by Dragos and magft
MickyMouse wrote: yes if u hardcode the indy to look at 10 mins, it will look there...
don't know about the trick with period converter, but im interested at what will show
It would be useful for back testing in the past beyond when we all started looking at Fratelli. I guess if buffer is made really big we can use ST to run the M1 chart and period converter to creat the M10 in the past. Once generated we can then run an EA to test best configuration, though to be honest it is pretty good already. I do use an extending TP and also a trailing stop so i can let profits run, still not convinced code is right. That is a job for the weekend :?

Right time for a Stella and start running some dynamic simulations of a gas lift system on an oil platform, oh the joys of deadlines at work :roll:

Micky, thanks for the help.
Author:  MickyMouse [ Fri Dec 02, 2011 9:40 pm ]
Post subject:  Re: Fratelli EA by Dragos and magft
yeap Stellaaaaaaaaaaaaa
Author:  kenfen [ Fri Dec 02, 2011 11:21 pm ]
Post subject:  Re: Fratelli EA by Dragos and magft
MickyMouse wrote:yeap Stellaaaaaaaaaaaaa
Hi Micky,

I have all those horrible GB's of Dukas raw tick data...and have spent FAR more hours than I care to remember doing 99% tick data backtests. :)

I've been following along with Fratelli. If you want a hand at all, I'll help in whatever way I can. I'm trying to get my head around a few things, though, because I'd like us to be damn sure the tests are going to be accurate before we commit the time and resources.

I follow along with kicking Empty4's dirty, fat arse into reading the indicators from M10 (nicely done!), but...when you run a backtest, it still must be reading from an .fxt file, no? So, logical choices would be M5 or M15, but M10 would be (I hate to say the word)...nearly impossible, no?

So...I believe 99% quality would be out, which is fine...just again trying to get my head around what we can expect to achieve. I guess the procedure would be...

-Convert the tick data and generate an M5 or M15 FXT file AND full History (.hst) files, too.
-Period Convert that fresh virgin M1 HST file to M10.
-Run backtests on M5 or M15...calling indicators from the M10 data.

Does that sound right?

Actually...the initial FXT file won't even matter, I believe. We would really be concerned with the fresh HST from the raw tick data. Get M10 from the fresh M1...and let Empty4 generate a new FXT file from the freshly created HST. Confused? :o

I have ALL of the data from Dukas...well, up until about a month ago...I'll grab the rest right now. That would give us about 4/5 years worth of tasty history.

Let me know your thoughts, mate.

All the best,

Ken

ps- I also have ALL of the OHLC candle data from Dukascopy...probably +10 years worth...but have always had some sort of problem when trying to use it/import it into Empty4; mis-match errors, or something. If anyone has experience with using that data...I have it all.
Author:  garyfritz [ Fri Dec 02, 2011 11:31 pm ]
Post subject:  Re: Fratelli EA by Dragos and magft
I have been swamped and am likely to stay that way for a while... but I'd really like to write a Tradestation strategy for Fratelli. That would give us a good reliable backtest and independent validation with another platform / another implementation. Will let you know if I get a few spare neurons to throw at that project...
Author:  kenfen [ Fri Dec 02, 2011 11:39 pm ]
Post subject:  Re: Fratelli EA by Dragos and magft
I REALLY have to hone my JForex skills!

I trade with Dukas live and have always threatened to polish all my weaknesses there. Ever since the Empty4 to JForex Bridge appeared...I keep kicking that can down the road...like a European politician! :twisted: :lol:

I'll see if I can make any progress there, too.
Author:  fratelli [ Sat Dec 03, 2011 12:15 am ]
Post subject:  Re: Fratelli EA by Dragos and magft
Hello
I see its here allthe coders are hiding.

Just want say hi

and keep up the good work.

fratelli
Author:  magft [ Sat Dec 03, 2011 12:48 am ]
Post subject:  Re: Fratelli EA by Dragos and magft
fratelli wrote:Hello
I see its here allthe coders are hiding.

Just want say hi

and keep up the good work.

fratelli
Nice of you to join us :lol:

I really like the system and it has a lot of potential. I think between all these great minds we should be able t come up with a solution to the testing and in the mean time forward testing is giving data to use.

Mike
Author:  kenfen [ Sat Dec 03, 2011 1:05 am ]
Post subject:  Re: Fratelli EA by Dragos and magft
The Creator is in the house...look sharp boys! :twisted: :lol:

@ Micky...

I now have ALL tick data up to today's close for ALL available pairs. Have a think about what I said a few posts above and let me know if you want to at least give it a shot to see what sort of results we get. If it looks promising, I have a machine I can dedicate to spin backtests around the clock to test ALL pairs.

AND if you wanted to try any other indicators in place of SuperTrend, this would be a great way to do it. I still think I have a couple of nice solutions. The results will be slightly different, but from what I've "eye-balled" maybe even better. It may prevent a couple of stinkers along the way.

Stellllllllla! :lol:
Author:  magft [ Sat Dec 03, 2011 1:49 am ]
Post subject:  Re: Fratelli EA by Dragos and magft
kenfen wrote:The Creator is in the house...look sharp boys! :twisted: :lol:

@ Micky...

I now have ALL tick data up to today's close for ALL available pairs. Have a think about what I said a few posts above and let me know if you want to at least give it a shot to see what sort of results we get. If it looks promising, I have a machine I can dedicate to spin backtests around the clock to test ALL pairs.

AND if you wanted to try any other indicators in place of SuperTrend, this would be a great way to do it. I still think I have a couple of nice solutions. The results will be slightly different, but from what I've "eye-balled" maybe even better. It may prevent a couple of stinkers along the way.

Stellllllllla! :lol:
Sounds interesting and will deinfately help the cause Viva La Stella :lol:

The attached version of the EA is for testing purposes only, i have been through and replaced all references to the current chart to with ChartTf which is defined as 10 at top of code. This should allows us to get the M10 values in ST using another tf like M5 or M15. I noticed that the period_converter_opt_mod updates the M10 history file on updates to the base tf chart ie if you got to M1 chart with indy on and press PageUp you get more M1 history and the M10 updates accordingly. So if you import tick data to M1 chart and then drop indy on it you get detailed M10 data to test out Micky's indy or the EA or both if you so wish.

Hope this helps, i'm off to bed now.

Mike
All times are UTC Page 16 of 23

Powered by phpBB® Forum Software © phpBB Limited