Jag's Phantom 6 System | Page 13 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Phantom 6 System 1,540 replies

JAG Trade Journal 0 replies

JAG's Journal 0 replies

Phantom high, phantom low 0 replies

JAG fx Trade Journal 10 replies

Attachments: Jag's Phantom 6 System
Exit Attachments

Jag's Phantom 6 System

  • #241
  • Jan 2, 2011 5:16am Jan 2, 2011 5:16am
You can compile the java file using the JForex platform... I expect you can google and figure it out its not that hard... I don't have time to write a tutorial on how to get it to work.

Backtesting.... it works well withink well defined constraints but we all know the market can do anything anytime.. so during any crisis period it will mess up because of ranging periods followed by trending that create a perfect storm.... I'm sure you can curve fit settings that work like they are doing inthe other threads... and create some kind of expectancy..aslong as the market allows them to keep doing it.

But in the long run you can't corner the market into giving you what you want... since the EA just jumps in without any strategy it loses starting from the first trade.

What you can do is apply some R:R.. What I meant by R:R is the amount your willing to win vs lose... By applying say 1% risk to 5% reward you can create a positive equity curve aslong as the system wins atleast 30% of the time.... now if your EA does a good job of detecting trends or getting inat the right times.. you can hopefully create some positive expectancy over a long period of time that will let you profit... however not doing this... you will get into a situation where you loss 50% before you make 0.4%..

The system is like martingale as is... so if you treat it that way.. you can set it up so that it may double your account and you can ride the rest for free. That is the only way to treat this strategy without applying hard risk reward cutoff ratio.

Jag
  • #242
  • Jan 2, 2011 5:56am Jan 2, 2011 5:56am
Thanks for the detailed explanation.
He was looking for the card so high and wild hed never need to deal another
  • #243
  • Jan 2, 2011 6:51am Jan 2, 2011 6:51am
thank you sidhujag

Why do this program in java?
it would not be easy to do MQL4?
Maybe you have the code in MQL4?
  • #244
  • Jan 2, 2011 11:16am Jan 2, 2011 11:16am
I tried testing this in JForex but although it ran it as though it was testing OK, I didn't get a single trade. I downloaded the last 3 months of all the pairs before I ran it. Naturally you only seem to be able to run the strat on one pair so I dont know if that has any relevance.
I also tried in visual mode and all the bars painted in as though live but still no trades.
I'm sure it must be something I'm doing thats wrong.
Any ideas would be helpful.
He was looking for the card so high and wild hed never need to deal another
  • #245
  • Jan 2, 2011 6:21pm Jan 2, 2011 6:21pm
You must select all 14 pairs in the historical data section for strategy tester.

Did you do that? Try to run in visual mode to see what the PL is for the phantoms.

How do you figure it can only run on one pair? When you do this it gets ticks and bars for every pair... this is different than mt4.
  • #246
  • Jan 2, 2011 6:22pm Jan 2, 2011 6:22pm
Quoting panthere
thank you sidhujag

Why do this program in java?
it would not be easy to do MQL4?
Maybe you have the code in MQL4?
Please read my posts
  • #247
  • Jan 3, 2011 1:48am Jan 3, 2011 1:48am
Quoting panthere
thank you sidhujag

Why do this program in java?
it would not be easy to do MQL4?
Maybe you have the code in MQL4?
Code is already in mql but you cannot backtest in mql that is why he migrated over to JTrader
  • #248
  • Edited 1:59am Jan 3, 2011 1:49am | Edited 1:59am
Quoting sidhujag
Hi guys,

here is the code let me know if it trades as its supposed to.

I tested on 200:1 @ 50k.

It should emulate AddSix_03bx_CL_TH_PS
Awesome thanks jag

Complicated code eish!
  • #249
  • Jan 3, 2011 2:51am Jan 3, 2011 2:51am
thank you very much for your explanations sidhujag
cordially
  • #250
  • Jan 3, 2011 3:22am Jan 3, 2011 3:22am
Quoting sidhujag
You must select all 14 pairs in the historical data section for strategy tester.

Did you do that? Try to run in visual mode to see what the PL is for the phantoms.

How do you figure it can only run on one pair? When you do this it gets ticks and bars for every pair... this is different than mt4.
My fault. I didn't realise the '+' button was there to add instruments.....They dont appear to have much documentation on how to use the backtesting functionality.

But its going now.....

While I'm here I must say a big thank you to you for all the trouble you've taken to re-code this for JForex.
He was looking for the card so high and wild hed never need to deal another
  • #251
  • Jan 3, 2011 4:26am Jan 3, 2011 4:26am
Quoting dr_who
My fault. I didn't realise the '+' button was there to add instruments.....They dont appear to have much documentation on how to use the backtesting functionality.

But its going now.....

While I'm here I must say a big thank you to you for all the trouble you've taken to re-code this for JForex.

You're telling me... if you ask me the platform sucks just because of the lack of docs... you spend 10k hours developing a java platform and you use some stupid javadocs with minimal explanation? Hire a tech writer and write some proper shiit with all the money your scamming off of your clients
  • #252
  • Jan 3, 2011 4:31am Jan 3, 2011 4:31am
Quoting Saidar
Code is already in mql but you cannot backtest in mql that is why he migrated over to JTrader

Not really...

Basically each currency calls an OnTick() for each tick and OnBar() for end of each bar... then OnAccount() when your equity changes. From there you can get an idea how to code a strategy.

The OHLC aren't good to test because sometimes you will close out your LONG at the HIGH while in reality you wouldn't have because you don't know where the high will be... thus higher profits in backtesting with same size losses.
  • #253
  • Jan 3, 2011 6:50am Jan 3, 2011 6:50am
Just finished my first historic test which took around 3 hours.
Tested last 3 months data.
Sadly the Dukascopy report isn't quite what I'd hoped which could have included drawdowns, %age P/Ls etc like we're used to in MT4.

Anyway, the big numbers are that starting from an account of $50K I ended up with a net profit of $4400 having turned over $11m and being charged £233 commission.

Jag, do you know a way of collatting the html report to make greater sense of it ? Its got all the individual trades so it should be possible with the right parser. Or do they make a collatted report available elsewhere ?

So, looking good so far, though I don't have details of biggest drawdown or %age margin used etc which would be handy. On observation it looked like the account never went below $40K at any stage but I'd like to get my hands on the above figures if possible.

I'll do another test using different dates.
He was looking for the card so high and wild hed never need to deal another
  • #254
  • Jan 3, 2011 7:12am Jan 3, 2011 7:12am
Sorry guys, hate to butt in but I thought i'd let you know that liteforex offer MT5 and it DOES allow you to backtest multiple pairs.

I know nothing about MT5. In fact i know nothing about MT4! But thought if i could offer some help towards your testing issues, then I would.

Ps - Awsome work.
  • #255
  • Jan 3, 2011 7:48am Jan 3, 2011 7:48am
Thanks for the info Viv. V useful.
He was looking for the card so high and wild hed never need to deal another
  • #256
  • Jan 3, 2011 9:08am Jan 3, 2011 9:08am
Quoting Viv108
Sorry guys, hate to butt in but I thought i'd let you know that liteforex offer MT5 and it DOES allow you to backtest multiple pairs.

I know nothing about MT5. In fact i know nothing about MT4! But thought if i could offer some help towards your testing issues, then I would.

Ps - Awsome work.
MT5 multi currency backtesting is OK at best. JForex is much more accurate.

The problem with MT5 is you cannot handle the tick of each currency pair separately, and there are huge synchronization problems when using bars instead of ticks.
  • #257
  • Jan 3, 2011 9:11am Jan 3, 2011 9:11am
Quoting sidhujag
Not really...

Basically each currency calls an OnTick() for each tick and OnBar() for end of each bar... then OnAccount() when your equity changes. From there you can get an idea how to code a strategy.

The OHLC aren't good to test because sometimes you will close out your LONG at the HIGH while in reality you wouldn't have because you don't know where the high will be... thus higher profits in backtesting with same size losses.
That is why you test with every tick? I know it takes forever but at least the results are accurate. Not much optimization is needed for this strategy though so not really a problem.

My version of the AddSix EA is completed will backtest it and post here if without problems. It is not nearly as complicated as your code though but it does not need to be as it is the basic version.

Still could not test your JForex strategy data still downloading it takes forever! Three hours left
  • #258
  • Edited 10:33am Jan 3, 2011 9:23am | Edited 10:33am
As you said yourself tho Saidar, its best to be accurate. There's no point in testing if the accuracy is in doubt.

Just started another 3 month test with earlier dates.

Jag, d'you know if there's any basket money management EA like Swiss Army Knife around ?
He was looking for the card so high and wild hed never need to deal another
  • #259
  • Jan 3, 2011 9:38am Jan 3, 2011 9:38am
Quoting dr_who
As you said yourself tho Saidar, its best to be accurate. There's no point in testing if the accuracy is in doubt.
Yes I don't think MT is even capable of using historical tick data. Although interpolation aint that bad...

The main problem like I said with MT5 is the multi currency backtesting is flawed to the core, something not right there!
  • #260
  • Jan 3, 2011 11:20am Jan 3, 2011 11:20am
Quoting dr_who

Jag, d'you know if there's any basket money management EA like Swiss Army Knife around ?

Found one coded by Steve Hopwood
He was looking for the card so high and wild hed never need to deal another
0 traders viewing now