Net Trap EA - variant backtesting

This is a discussion on Net Trap EA - variant backtesting within the Forex Net Trap forums, part of the T2W Archive category; Originally Posted by surferking Thanks, ok, that's one step closer, now it takes a long time but still doesn't trade. ...

Reply
 
LinkBack Thread Tools Search this Thread
Old Mar 22, 2011, 9:54pm   #91
 
surferking's Avatar
Joined Oct 2010
Re: Net Trap EA - variant backtesting

Quote:
Originally Posted by surferking View Post
Thanks, ok, that's one step closer, now it takes a long time but still doesn't trade.

I get the message "trade error [134] not enough money".
As a quick fix putting 1000000 as the starting bank get it going but I wanted to stick to the 10000 to keep in line with other tests.
__________________

Tick follows tock follows tick.... http://www.youtube.com/watch?v=EJeXQniTKto
surferking is offline   Reply With Quote
Old Mar 22, 2011, 10:24pm   #92
 
surferking's Avatar
Joined Oct 2010
Re: Net Trap EA - variant backtesting

Strangely it seems all the prices have an additional 3 decimal places / zeroes on the end so maybe something peculiar about the Dukascopy tick data. (see screenshot)

I followed the instructions from here to the letter: http://eareview.net/tick-data

Any ideas? Ideally I'd like to run everything like-for-like with others for the most useful comparison.
Attached Thumbnails
screen-shot-2011-03-22-22-march-22.14.40.png  
__________________

Tick follows tock follows tick.... http://www.youtube.com/watch?v=EJeXQniTKto
surferking is offline   Reply With Quote
Old Mar 22, 2011, 11:50pm   #93
 
surferking's Avatar
Joined Oct 2010
Re: Net Trap EA - variant backtesting

Here's a better screenshot of the MT Test Journal Log Entries and the tester log file if it helps...
Attached Thumbnails
screen-shot-2011-03-22-22-mar-2011-23.46.59.png  
Attached Files
File Type: zip 20110322.log.zip (29.4 KB, 17 views)
__________________

Tick follows tock follows tick.... http://www.youtube.com/watch?v=EJeXQniTKto
surferking is offline   Reply With Quote
Old Mar 22, 2011, 11:56pm   #94
 
surferking's Avatar
Joined Oct 2010
Re: Net Trap EA - variant backtesting

Sorry to clutter this thread but here's a screenshot of a slightly different error I get when trying the FNT EA v008.
Attached Thumbnails
screen-shot-2011-03-22-22-mar-2011-23.53.22.png  
__________________

Tick follows tock follows tick.... http://www.youtube.com/watch?v=EJeXQniTKto
surferking is offline   Reply With Quote
Old Mar 23, 2011, 12:04am   #95
 
surferking's Avatar
Joined Oct 2010
Re: Net Trap EA - variant backtesting

Perhaps something to do with this odd large negative FreeMargin value?
Attached Thumbnails
screen-shot-2011-03-23-23-mar-2011-00.03.04.png  
__________________

Tick follows tock follows tick.... http://www.youtube.com/watch?v=EJeXQniTKto
surferking is offline   Reply With Quote
Old Mar 23, 2011, 12:42pm   #96
 
MaxiV's Avatar
Joined Dec 2010
Re: Net Trap EA - variant backtesting

Quote:
Originally Posted by surferking View Post
Hey Maxi,

I'm having trouble back-testing this EA. I've downloaded the Dukascopy data following the instructions from here to the letter to install it: http://eareview.net/tick-data

However, when I run the back test it only seems to trade one tick then it finishes.

Don't suppose you've come across this before or have any ideas?
I'm having truble now doing EUR backtest... just doesn't want to do it... cant see whats wrong.... you want to speak to SpinDrift on this... he's the MT expert...

Can't see why after i finished GBP (one graph not fully complete)... it didn't want to do any more. No change in data didn't do any changes??? don't know!??

Someone help us both.
__________________
Trade like a Samurai: with discipline but without emotion....
http://twitter.com/MMVaghela
MaxiV is offline   Reply With Quote
Old Mar 23, 2011, 1:36pm   #97
 
MaxiV's Avatar
Joined Dec 2010
Re: Net Trap EA - variant backtesting

Quote:
Originally Posted by MaxiV View Post
I'm having truble now doing EUR backtest... just doesn't want to do it... cant see whats wrong.... you want to speak to SpinDrift on this... he's the MT expert...

Can't see why after i finished GBP (one graph not fully complete)... it didn't want to do any more. No change in data didn't do any changes??? don't know!??

Someone help us both.
Hmmm,,, got it working again, re exported the data i had and did the period script thing. now finishing off the 2009 MMI AI-RT to complete and add.


Will this same method work for the EUR/USD? Just tried...nothing.... what do i need the change (is it because its in GBP)?
Will see how accurate this is....
__________________
Trade like a Samurai: with discipline but without emotion....
http://twitter.com/MMVaghela

Last edited by MaxiV; Mar 23, 2011 at 4:25pm.
MaxiV is offline   Reply With Quote
Old Mar 24, 2011, 4:32pm   #98
 
MaxiV's Avatar
Joined Dec 2010
Re: Net Trap EA - variant backtesting

Quote:
Originally Posted by MaxiV View Post
Ok

Here is the back test on MMI AI-RT v4 : From 2005-2008 (you all should have 2010 backtest already). Again note; 2006 was not a great year.

Starting balance 10000 for each year @ 3%MM compounded.

Please Note.... 2009 my PC decided to pack up !!! So 2009 is only upto Aug... my appologies!

However, the full compounding from 2005 to 2008 (four years) is compleate for each and will post later today.
Here's the fully completed 2009 chart for MMI AI-RT (which was incomplete).

I now want to do EUR/USD but not sure what settings i need to change. Anyone tell me and i can get the backtesting done for EUR as I did for GBP/USD.
Attached Thumbnails
strategytestermmiai30comp2009.gif  
__________________
Trade like a Samurai: with discipline but without emotion....
http://twitter.com/MMVaghela

Last edited by MaxiV; Mar 24, 2011 at 4:40pm.
MaxiV is offline   Reply With Quote
Old Mar 25, 2011, 9:00pm   #99
 
surferking's Avatar
Joined Oct 2010
Re: Net Trap EA - variant backtesting

Attached is my first test run using Dukascopy tick data converted to M1 bars. (I'm having trouble getting MT4 to recognise the tick data .fxt implanted file at the moment so using the M1 .hst files to generate the tests, which should be accurate to 90% anyway).

This is just a graph and report on £10k MMI AI RT 30INT30 with 5%MM compounded from 1/1/2010 to 11/3/2011, which takes you to a balance of £168k over the last 14.5 months. (Parameter details removed in respect of their various authors.)

Sound good until you consider being up and down by £8k per day! That would take some big clanging brass balls to keep your cool with, and judging but the daily 'discretion' (meddling) on the board I'm sure there'd be some serious psychology to overcome to stick to that game plan. (and a max drawdown £62k, ouch)

Its just a test of my Tester setup so I'll work on more methodical set of tests later. I'm still trying to debug an issue so I can truly test with 99% quality.

The most obvious thing I can see from all these initial tests is that trading December and the 1st week of January doesn't seem to have any net benefit over the last 5 years so I won't be trading those dates this year. Summer and other holidays don't seem significant but that's just my general opinion.

I'm going to repeat setup and date range on a few different pairs first though to see how they work to satisfy a bit of curiosity.

I'm also considering trying JPY pairs from 10pm to 12am in the 2 hours before Tokyo opens to see if there's a similar behaviour there.

Out of interest, 1 year of testing is taking about 35 minutes to run on my VM (then the remaining 3 months took another 30 mins, probably dues lots of calculations for unused options). I don't think MT4 is multi-processor aware so I'm sure MT5, whenever that becomes mainstream, will improve times.

I'm not taking requests yet as I have a queue of my own back-tests to do as well as getting 2003-2007 data prepared, but any ideas I might look at later.

MaxiV If you want to send me a .set file and date range to do any direct comparisons to get an idea of the data integrity then ping it across.
Attached Thumbnails
gbpusd-mmi-ai-rt-25lockin-30int30-20100101-20110311strategytester.gif  
Attached Files
File Type: zip GBPUSD-MMI-AI-RT+25lockin-30INT30-20100101-20110311StrategyTester.zip (27.8 KB, 22 views)
__________________

Tick follows tock follows tick.... http://www.youtube.com/watch?v=EJeXQniTKto
surferking is offline   Reply With Quote
Old Mar 26, 2011, 9:31pm   #100
 
surferking's Avatar
Joined Oct 2010
Re: Net Trap EA - variant backtesting

OK, after a lot of trial and error, head banging and endless questions to Spin and Birt I now have Dukascopy Tick Data 99% Quality Back Testing running from 2007 to date.

I followed the instructions from here to the letter: http://eareview.net/tick-data and the problem seems to be that the .fxt files don't bl**dy work in MT4 on SLM!

On the off chance, and because I wanted to test something else anyway, I installed MT4 Demo for FXOpen ECN and the data worked first time.

So, because it doesn't matter who's MT4 I use since the data is coming from Dukascopy, I'm now in action for 99% quality Tick Data backtesting.

I'm going to repeat a few of the ones that MaxiV has done to compare results and see if there's much difference, then I'll probably test Goldfinger's 5INT40 because I believe tick data is the only reasonably reliable way to properly test such a scalping strategy on M15 bars.

I have 15 different pairs of tick data so I'm going to work my way through testing those on the most popular variants to see if any other pairs are worth diversifying with, then I'm going look at some experimental stuff and try and finish off a Newsnight Trader EA and back-test that because I have some filters and improvements in mind I need to evaluate.

Anything interesting I'll post here....
__________________

Tick follows tock follows tick.... http://www.youtube.com/watch?v=EJeXQniTKto
surferking is offline   Reply With Quote
Old Mar 26, 2011, 10:03pm   #101
 
surferking's Avatar
Joined Oct 2010
Re: Net Trap EA - variant backtesting

MaxiV and anyone else backtesting - I can highly recommend the "Loader" on Birt's page: http://eareview.net/tick-data

You find the right one for the build of MT4 you're running and then drop it in the same folder as the terminal.exe but then double-click on the loader.exe instead to have many MT4 instances running with all the same settings and data so that you can run multiple back-tests in parallel. Very handy!

(oh, and the loader doesn't seem to work on SLM either, I'm wondering if their SB customisation makes it incompatible with all these hacks.)
__________________

Tick follows tock follows tick.... http://www.youtube.com/watch?v=EJeXQniTKto
surferking is offline   Reply With Quote
Old Mar 26, 2011, 11:46pm   #102
 
surferking's Avatar
Joined Oct 2010
Re: Net Trap EA - variant backtesting

OK, initial results on EURUSD and AUDJPY as alternate tests using MMI AI RT 30INT30 +25 lockin doesn't look very promising.

Does anyone want to suggest a good 'middle of the 'road' variant that I can test across all 15 pairs that I have tick data for to highlight / eliminate once and for all if this strategy can be used on other pairs?
__________________

Tick follows tock follows tick.... http://www.youtube.com/watch?v=EJeXQniTKto
surferking is offline   Reply With Quote
Old Mar 27, 2011, 4:00pm   #103
 
MaxiV's Avatar
Joined Dec 2010
Re: Net Trap EA - variant backtesting

Quote:
Originally Posted by surferking View Post
Attached is my first test run using Dukascopy tick data converted to M1 bars. (I'm having trouble getting MT4 to recognise the tick data .fxt implanted file at the moment so using the M1 .hst files to generate the tests, which should be accurate to 90% anyway).

This is just a graph and report on £10k MMI AI RT 30INT30 with 5%MM compounded from 1/1/2010 to 11/3/2011, which takes you to a balance of £168k over the last 14.5 months. (Parameter details removed in respect of their various authors.)

Sound good until you consider being up and down by £8k per day! That would take some big clanging brass balls to keep your cool with, and judging but the daily 'discretion' (meddling) on the board I'm sure there'd be some serious psychology to overcome to stick to that game plan. (and a max drawdown £62k, ouch)

Its just a test of my Tester setup so I'll work on more methodical set of tests later. I'm still trying to debug an issue so I can truly test with 99% quality.

The most obvious thing I can see from all these initial tests is that trading December and the 1st week of January doesn't seem to have any net benefit over the last 5 years so I won't be trading those dates this year. Summer and other holidays don't seem significant but that's just my general opinion.

I'm going to repeat setup and date range on a few different pairs first though to see how they work to satisfy a bit of curiosity.

I'm also considering trying JPY pairs from 10pm to 12am in the 2 hours before Tokyo opens to see if there's a similar behaviour there.

Out of interest, 1 year of testing is taking about 35 minutes to run on my VM (then the remaining 3 months took another 30 mins, probably dues lots of calculations for unused options). I don't think MT4 is multi-processor aware so I'm sure MT5, whenever that becomes mainstream, will improve times.

I'm not taking requests yet as I have a queue of my own back-tests to do as well as getting 2003-2007 data prepared, but any ideas I might look at later.

MaxiV If you want to send me a .set file and date range to do any direct comparisons to get an idea of the data integrity then ping it across.
If you would... im just wondering what you get for the whole of 2006, as you may know from my backtest gave a loss year but not for MMI AI-RT at 3% (however, have been working on backtesing the whole 2005-2008 @ 5% but thanks to a power faluar the data was lost and MT4 does have no smart auto save!... but i did do a save this morning and let it carry one... sooo close to the end was at 2008 September!). As you can see 5% is great if, which i do, have confidence in this system.

Also, trying to do the EUR/GBP... but cant seem to get it working.
Attached Thumbnails
strategytestermmi5-mmcomp2005-2008.gif  
__________________
Trade like a Samurai: with discipline but without emotion....
http://twitter.com/MMVaghela
MaxiV is offline   Reply With Quote
Old Mar 27, 2011, 4:00pm   #104
 
mark120169's Avatar
Joined Apr 2010
Re: Net Trap EA - variant backtesting

Quote:
Originally Posted by surferking View Post
OK, initial results on EURUSD and AUDJPY as alternate tests using MMI AI RT 30INT30 +25 lockin doesn't look very promising.

Does anyone want to suggest a good 'middle of the 'road' variant that I can test across all 15 pairs that I have tick data for to highlight / eliminate once and for all if this strategy can be used on other pairs?
I am confused why the search for "other pairs" ? As the GU spread with SLM is 2 pips, only one pair is better, EU @ 1.5 pips, and a few the same 2 pips, every other pair the spread is worse.

Also a large percentage of systems run on GU becasue of the daily movements that it creates.

So i'm not sure why with the good spread and daily movement it is nesscessary to find other pairs if GU is working ok, which it is for us now?

2011 is profitable with our tweaks, 2010 was profitable (by almost 100%) and it seems with the backtesting only 2006 was a bad year but even MMI AI RT turns that into a small profit.

What more do you need? Surely just compound the stakes weekly with our tweaks and keep at it.

M
mark120169 is offline   Reply With Quote
Old Mar 27, 2011, 4:25pm   #105
 
surferking's Avatar
Joined Oct 2010
Re: Net Trap EA - variant backtesting

Quote:
Originally Posted by MaxiV View Post
If you would... im just wondering what you get for the whole of 2006, as you may know from my backtest gave a loss year but not for MMI AI-RT at 3% (however, have been working on backtesing the whole 2005-2008 @ 5% but thanks to a power faluar the data was lost and MT4 does have no smart auto save!... but i did do a save this morning and let it carry one... sooo close to the end was at 2008 September!). As you can see 5% is great if, which i do, have confidence in this system.

Also, trying to do the EUR/GBP... but cant seem to get it working.
Sorry, the Dukascopy tick data only starts from 30-03-2007. I might be able to try their M1 data but haven't got that far yet. I hear Oanda can provide tick data back to 2003 if you have $1,000 on deposit with them so that will be my next step. I'll get all the 2007-2001 tests I have in mind done first though because recent history is probably more applicable for most anyway.
__________________

Tick follows tock follows tick.... http://www.youtube.com/watch?v=EJeXQniTKto
surferking is offline   Reply With Quote
Reply

Thread Tools Search this Thread
Search this Thread:

Advanced Search

Similar Threads
Thread Thread Starter Forum Replies Last Post
Net Trap Variant live results from 17/05/10 mark120169 Forex Net Trap 572 Dec 5, 2011 10:54pm
Net Trap Tools mark120169 Forex Net Trap 10 Dec 11, 2010 9:02pm
Forex Net Trap SA Trader Forex Net Trap 6 Jun 28, 2010 1:44pm
Net Trap Johnnysportster First Steps 5 Mar 30, 2010 9:08am

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)