Art Phantom EA | Page 32 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Phantom 6 System 1,540 replies

Phantom 6 - Manual System 8 replies

Phantom Entry on ODL Japan. 2 replies

Tradeview CAD/JPY "Phantom" spike 0 replies

Phantom high, phantom low 0 replies

Attachments: Art Phantom EA
Exit Attachments

Art Phantom EA

  • #621
  • Nov 28, 2010 6:58am Nov 28, 2010 6:58am
Quoting roundrock
i rewrote the code so that we can backtest. this is a single pair phantom strategy, you can drop to any chart any timeframe. it will trade that pair and that timeframe. if you drop it on EU 1 Hr chart, then it will trade EU pair and adds are done on start of the hour. if you drop it on 30min chart then it trades every 30 min at start of bar

when you start EA , it remembers the current price. then at start of the next bar, if the pair has moved up (with respect to start price), and if all prev Long trades are in profit then it will add a Long....
roundrock,
We won't be able to use this single phantom unless we are allowed to hedge, right? thanks
  • #622
  • Nov 28, 2010 7:10am Nov 28, 2010 7:10am
Quoting Toothman
roundrock,
We won't be able to use this single phantom unless we are allowed to hedge, right? thanks
I think you are right but if it works, there could be a solution for non allowed hedge brokers......
  • #623
  • Nov 28, 2010 7:30am Nov 28, 2010 7:30am
Quoting Caillou
I think you are right but if it works, there could be a solution for non allowed hedge brokers......
Am allmost done with backtest 30 min from jan till now with time filter :
+ 30% , max DD about 20% (this is aprox, looking at strateGy tester)
PP is 0.5
  • #624
  • Nov 28, 2010 8:53am Nov 28, 2010 8:53am
Roundrock or someone else

Can you please point out where exactly your latest version for single pair is placed in this thread?
Well I am looking forward for two pairs EU and UC with same method.

Thanx in advance.
  • #625
  • Nov 28, 2010 10:40am Nov 28, 2010 10:40am
Quoting Caillou
What is exactly your problem?
I load the 'EA on a chart (eg GBP / USD), and then what? He manages all the pairs ? Or should I load the EA on 3 graphics? I do not understand the logic of how it is handled basket.

Roberto
  • #626
  • Nov 28, 2010 12:22pm Nov 28, 2010 12:22pm
Quoting 24hrs
Roundrock or someone else

Can you please point out where exactly your latest version for single pair is placed in this thread?
Well I am looking forward for two pairs EU and UC with same method.

Thanx in advance.
You can see attachments on the top right of the page.
  • #627
  • Nov 28, 2010 12:24pm Nov 28, 2010 12:24pm
Quoting antani
I load the 'EA on a chart (eg GBP / USD), and then what? He manages all the pairs ? Or should I load the EA on 3 graphics? I do not understand the logic of how it is handled basket.

Roberto
What EA? Please read the thread, I'm sure you will find your answers
  • #628
  • Nov 28, 2010 12:32pm Nov 28, 2010 12:32pm
Hey vienna! I'm glad to see you here

Quoting vienna
hello,Caillou,roundrock!
Please do not annoy,
they both do a super work-
.....
we stay cool and think of the ultimate goal making money!
Best Best regards
vienna
Charts never lie! (everyone else does) ... now where's that FF Adult lounge
  • #629
  • Nov 28, 2010 12:42pm Nov 28, 2010 12:42pm
Hi Caillou,
I think vienna's suggestion is a good one.
You may include me if you like. I have 8 computers at my disposal all running 24-7 of course I already am testing several versions now but only running a couple days so need more time before I post about them.

I feel one problem is the versions are updating so fast it's hard to know which ones to continue with and which ones to kick in favor of starting the newer one. While I'm not at all against quick version updates, if you have the mind to instruct: 1) which to test, 2) for how long and possibly even 3) the settings you'd like to see tested out to us, I think that could help alot. Just MHO is all.

TJ

Quoting Caillou
Absolutely agree, next week I will organize that...

Thank you
Charts never lie! (everyone else does) ... now where's that FF Adult lounge
  • #630
  • Nov 28, 2010 12:48pm Nov 28, 2010 12:48pm
May I suggest you post the backtest so we can see that it is viable.
Regards TJ

Quoting vanupips
My backtest without fiter:
30min E/U from jan till now 95% DD 70%
15 min looked better ,but I only succeeded the backtest for 6 month, sometrhing went wrong , does not work.
1H did not succeed in backtesting, got errors.
... suddenly going very slowly, I have no idea why
Charts never lie! (everyone else does) ... now where's that FF Adult lounge
  • #631
  • Nov 28, 2010 2:11pm Nov 28, 2010 2:11pm
Quoting TraderJoe
Hi Caillou,
I think vienna's suggestion is a good one.
You may include me if you like. I have 8 computers at my disposal all running 24-7 ...
Hi TraderJoe and everyone,

You're right. As I said before Vienna's suggestion is very interesting. The thread in only one week has gone extremely fast what is good but we have to make a calendar, in this way we will set goals and time to achieve them.

About the EA, please correct me if I'm wrong, I think "Single_Pair_Phantom" is the best candidate to test seriously for 2 reasons:

1.- In my opinion, the performance might be better with only one pair: the main advantage of this EA is the Hedge and with one pair it could be more efficient, less trades will be taken and less spread for each basket.

2.- With only one pair we could do backtests, a big advantage that we must use.

I have not rejected the 3 pair version but I think it is better, easy and quicker to test 1 pair version.

Our main goals would be:

1.- Set optimum MMTU and PP for a good Gain/DD.

2.- Set if needed some filters to minimize DD.

What I suggest is:

1.- Week 29/11-3/12:

a.- Trough Backtests we have to set Goal 1 and maybe we could find 2 optimum settings: aggressive and conservative, each one with a different MMTU and PP. For now I think most people have used MMTU:0.4 and PP: 1.0. I think 2 traders would be good to confirm results. We will call them Trader A and Trader B. We have to test EUR/USD, M30, for this month, this year and a more bigger time (we can decide this).

b.- Trough Forwardtest, we have to think about Goal 2. For this, a brainstorming could be interesting, then we can choose the best ones according to all our opinions. The more suggestions, the more possible potential.


2.- Week 6/12-10/12:

a.- If needed some filters, we could ask Roundrock to include in the EA the best winners modifications. One more time, thanks in advance.

b.- Trough Backtests, we have to test the new EA version with mods. for aggressive and conservative settings. I think the best in order to compare first and new version is that these backtests were made by Trader A and Trader B.


3.- From Week 13-12/17-12 till ???:

a.- We can make a Forward Tester Team: 50% with aggressive and 50% with conservative settings, the same day, at the same hour, with diferrents brokers, each member could start a demo in order to compare results and test the stability of the EA.


b.- After some weeks of test ( we have to fix how many weeks), the same Forward Tester Team could start a live trading with a very small account (or not, it will depend of each trader), in order to compare results between Demo and Live.


If we arrive to stage 3-b with good and stable results, I think we could feel proud. After that it depends of each trader.

As you suggested TraderJoe, I count on you.

So the first thing is to have Trader A and Trader B. Any voluntary? Of course, we can have 2, 3, or more, we only have to organize....
  • #632
  • Nov 28, 2010 4:07pm Nov 28, 2010 4:07pm
I think we need a spread filter.

My version just traded a EJ buy with a 24.5 pip spread!

Is there any way of tracking the normal spread for each pair besides going to your brokers page and looking up the typical spreads for each pair?
  • #633
  • Edited 6:46pm Nov 28, 2010 4:47pm | Edited 6:46pm
Pair: EURUSD
TF: M15

Starting Capital - $1000
Leverage - 500:1

MMTU - 0.4
PP - 1.0

*** I've added the PipStep = 10 mod in my own backtester as the only filter ***

Period: 02 Sep 2010 to 26 Nov 2010

M15
---
EDIT: After a second backtest, the account was STOPPED OUT LOL! FORGET IT!

******

Using the same settings, but on the following time frames:

M30
---
Last Fully Closed set of trades at 23 Nov 2010 17:20

91.25% Profit

H1
--
Last Fully Closed set of trades at 23 Nov 2010 18:25

73.87% Profit

H4
--
Last Fully Closed set of trades at 23 Nov 2010 18:12

18.44% Profit

D1
--
Last Fully Closed set of trades at 23 Nov 2010 03:06

17.39% Profit


So far, the results are telling me that...

The higher the timeframe the lower the DD (lower risk of margin call), but lower profit expectancy.

i.e. - M15 was a profit monster until the margin call.

Using that information, I would use Art_Phantom_Light as it gives good profit with lower DD than Art_Phantom_Full.
  • #634
  • Nov 28, 2010 4:47pm Nov 28, 2010 4:47pm
if you mean manual tracking? or built into the EA shown onto the chart?


Quoting the-game
I think we need a spread filter.

My version just traded a EJ buy with a 24.5 pip spread!

Is there any way of tracking the normal spread for each pair besides going to your brokers page and looking up the typical spreads for each pair?
Charts never lie! (everyone else does) ... now where's that FF Adult lounge
  • #635
  • Nov 28, 2010 4:51pm Nov 28, 2010 4:51pm
Quoting Caillou
Hi TraderJoe and everyone,

You're right. As I said before Vienna's suggestion is very interesting. The thread in only one week has gone extremely fast what is good but we have to make a calendar, in this way we will set goals and time to achieve them.

we only have to organize....
Hi Caillou... so we go with the 1 pair basket version till it is proven one way or the other, then if needed, we can pickup again with the 3 pair version and prove or disprove it.

In order to avoid assumption errors I suggest the full name of the EA each person decides to test be redundantly posted somewhere in every post where any kind of test or settings results are posted - No Accronyms! - there are just too many versons floating around to be able to keep clarity going within a team without doing this, IMHO.


Quoting Caillou
Week 29/11-3/12
---------- Initial Goals Accomplished by Backtesting ----------
1.- Set optimum MMTU and PP for a good Gain/DD [optimal and aggressive]
2.- Set if needed some filters to minimize DD....

Need (at least) 2 team members (maybe 4) to confirm Results for testing MMTU=0.4 PP=1.0 on two separate platforms.
I'm thinking we should also have some diversity of leverages for instance 500/1, 100/1 and or 50/1 in these tests too.
TRADER A MMTU= 0.4 PP=1.0 LEVERAGE=?
TRADER B MMTU= 0.4 PP=1.0 LEVERAGE=?

NOTE: We need a full (common fields) list of Settings to be tested before we can begin to hand them out.
Above are two i.e., [MMTU= 0.4 PP=1.0 LEVERAGE=] are their any others, if so what are they?

NOTE: Backtesting should only be used for the purposes of finding MMTU and PP settings close enough or "in the ballpark" so to speak for Forward testing purposes. This way we can save time getting close enough to forward test and find the MMTU and PP sweet spots.

On a Personal Note: I have had many hours behind the backtest wheel and my suggestion here is to only use 99% backtesting.
I am currently working on building (hopefully have it soon) an up to date (data-wise) 99% backtester (maybe someone else here already has one? Although I do currently have a 99% backtester with about 3 years of data but the data is not current and ends somewhere in 2007-2008 I believe. Nevertheless, it can still be used to dial-in an "in the ballpark" setting for MMTU and PP to forward test from there. If you'd like me to use it let me know meanwhile I'm still putting together the updated one.

Regarding the brainstorming. I'm thinking a skype room (or equivilent) or even a separate thread may be best for brainstorming since that type of activity might caues too much commotion and clutter in this area? Then the agree ideas could be brought into the thread. How do you feel about this?

We need to continually refer back to Caillou's post #631 as we continue to avoid having to quote such a large message everytime.

TJ
Charts never lie! (everyone else does) ... now where's that FF Adult lounge
  • #636
  • Nov 28, 2010 4:56pm Nov 28, 2010 4:56pm
Quoting the-game
Pair: EURUSD
TF: M15

Starting Capital - $1000
Leverage - 500:1

MMTU - 0.4
PP - 1.0

Period: 02 Sep 2010 13:30 to 24 Nov 2010 21:07

Net Profit: $1662.51 (166.25%)

******

Using the same settings, but on the following time frames:

M30
---
TBA

H1
--
TBA

H4
--
TBA

D1
--
TBA
166,25% ????
  • #637
  • Nov 28, 2010 4:59pm Nov 28, 2010 4:59pm
Caillou is looking for team members to take on some of the testing Caillou && RoundRock will be needing.

Those interested please reply with how many systems can you reasonably test at the same time.

My opinion is that one normal computer should be able to run at least 2 platforms if not 4. More is certainly doable but then again it does depend on that particular computers nuances and what you are currently already running on it.
---------
If you are not experienced at running EAs but would still like to help, I found a couple of tutorials on youtube which should help get you up to speed. I was going to post them but then realized posting these links may be against FF policy so just google "How do I run an EA on MT4" and you'll find plenty of help.

If you wish to be apart of the team please sound off and I'll make a list. Please keep the messages as brief as possible

TJ
Charts never lie! (everyone else does) ... now where's that FF Adult lounge
  • #638
  • Nov 28, 2010 5:03pm Nov 28, 2010 5:03pm
Quoting Caillou
166,25% ????
start with $1000 ended up with 2662.51, Net Profit is $1662.51, meaning Net Profit % is 166.25%

Doesn't matter anyway, my second backtest was STOPPED OUT, meaning my account had no margin to trade with.

I'm not sure how reliable the MT4 Strategy Tester is if it gave me two different results in the space of 15 mins between each test.
Maybe it downloaded more data? I don't know...

EDIT: Seems like when it catches a HUGE bull run, therefore will add many buy trades, then the trend suddenly goes short that's when you can experience margin call.

For now I'll be sticking to my Art Extreme with just the PipStep = 10 filter
  • #639
  • Nov 28, 2010 5:16pm Nov 28, 2010 5:16pm
My computer can easily manage 2 platforms running at the same time, 24/7. Would be glad to help with testing the 1 pair EA, let me know which EA/settings to test

Still, there's a couple of questions I still haven't gotten around: Somewhere along the thread sidhujag shared a code (based on the 6 pair EA's) that didn't rely on the global variables to work, but instead used a different type of calculation for the baskets (I have to mention that I am no coder).

From what I understood, this type of coding for the "phantom" EA worked better than the one we're using right now? Maybe (Probably) I'm mistaken, but can someone please elaborate a little about this specific issue? How often do we have to delete the global variables for the EA to work fine? I still haven't got this piece of the EA figured out.

Thanks beforehand and sorry if this question has already been answered before (I don't recall it though).
  • #640
  • Nov 28, 2010 5:19pm Nov 28, 2010 5:19pm
Hi roundrock,

The Trading Hours code you implemented in the backtester version of Single Pair Phantom does not use the StopAfterNoTrades flag, it simply just stops the EA from trading at all.

During the non-trade hours, the EA should continue to try and "finish" an existing sequence of trades or if there isn't any open trades, just don't trade at all.

Thanks!
0 traders viewing now