EA Factory | Page 2 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Calendar as Fundamental Source at Forex Factory 14 replies

Forex Factory Store 0 replies

Journal Factory posters - how much money are you making? 17 replies

Newbie Factory Journal 12 replies

Attachments: EA Factory
Exit Attachments

EA Factory

  • #21
  • Sep 29, 2006 8:45pm Sep 29, 2006 8:45pm
my only comment is that this is a beautiful thread, a testament to the strength and kindness of our community here.
Relax and be happy.
  • #22
  • Sep 29, 2006 9:24pm Sep 29, 2006 9:24pm
I took a look at last two orders executed by Mouteki EA on backtesting cable.

order605
- is opened correctly (or at least how I understood the method): the candle is under the TL.
- but the TP, I think it isn't projected right and is closed prematurely

order606
- is opened above the TL. Or EA did not redraw the new TL, and it's using the TL from order605????
Attached Images
The trend isyour friend untilis about to end...
  • #23
  • Sep 29, 2006 11:57pm Sep 29, 2006 11:57pm
Quoting aha
Here is the EA I made for Mouteki system. It is currently being tested by many members on this forum. You may downloaded here and either backtest it or forward test it.

You may test it silently, there is no obligation to post your findings (bug report, trade setup or results, ideas to modify, etc.) here. But if you would like to share your findings, please do it in an organized but convenient way.
hey AHA in testing your EA, i manually began going through the trades your EA made. i caught one trade that your EA missed (i think i caught it at least).

the trade should have been a sell on august 10, 2006, 4:00
at 1877.
Attached Image
  • #24
  • Sep 30, 2006 12:25am Sep 30, 2006 12:25am
Oops, you got me there.

The EA has been modified to take trades on the opening of the bar but I forgot to modify the comment lines.

I should thanks Attiqul here because it was him found this bug initially and notified me.


Quoting hans_gregor
Regarding your comments lines from Mouteki EA.
// Entry Conditions:
//
// 1. Enter long trade
// 1) Enter long when price close higher than the top trend line. The
// entry price is the first open price above the top trend line.
2. Enter short trade
// 1) Enter short when price close lower than the bottom trend line.
// The entry price is the first open price below the bottom trend
// line.

Reading Mouteki's document, I understood that to open a position you don't have to wait until the candle is closed. I don't know if your EA open in this way the positions, or not . Or if this issue has been clarified somewhere outside the original document (forum) by Mouteki.
Target: Trading naked.
  • #25
  • Sep 30, 2006 12:36am Sep 30, 2006 12:36am
Let me defence the TP rule for the order 605 first:

In the Mouteki system the TP is calculated based on the highest bar, not the bar will produce the largest TP. In the chart you posted, if we call the entry bar as bar 0, the EA actually followed the Mouteki system by calculating the TP using the bar 1, while you calculated the TP using bar 7.


Quoting hans_gregor
I took a look at last two orders executed by Mouteki EA on backtesting cable.

order605
- is opened correctly (or at least how I understood the method): the candle is under the TL.
- but the TP, I think it isn't projected right and is closed prematurely

order606
- is opened above the TL. Or EA did not redraw the new TL, and it's using the TL from order605????
Target: Trading naked.
  • #26
  • Edited 1:58am Sep 30, 2006 12:52am | Edited 1:58am
Hi Hans_gregor,

I forgot to thank you for posting these charts and question. Your contributions are important for developing high quality EAs.

As for order 606, I am sure the EA redrew TL before checking the setup. It's weird that it opens a trade like that because I double checked my code but didn't find out why. A small possibility here is, since the EA checks the setup in the first 10 seconds of each 4 hour, even though the opening price is higher than the TL, the price may have dropped below the TL in that 10 sec, but that's not likely.....

I should be able to find out what exactly happened there if you can help me debug this. Please PM me if you are interested.

[Edit] Wait a second, your black bars are bullish bars. So the trade was actually taken at the opening of the bar, I thought that was the close of the bar. Ok that explains it, since the opening price of the bar was very close to the TL so it's likely that in 10 second the price has dropped below the TL and trigger the EA to place the trade.

Maybe I should define an offset parameters here. A TL breakout is valid only when the price breaks the TL by a certain offset (say 2 pips). But that may produce other missing trades, maybe not a good idea. Let's just keep it unchanged for now.

aha

Quoting hans_gregor
I took a look at last two orders executed by Mouteki EA on backtesting cable.

order605
- is opened correctly (or at least how I understood the method): the candle is under the TL.
- but the TP, I think it isn't projected right and is closed prematurely

order606
- is opened above the TL. Or EA did not redraw the new TL, and it's using the TL from order605????
Target: Trading naked.
  • #27
  • Edited 2:30am Sep 30, 2006 12:55am | Edited 2:30am
Hi Phoenix,

Thank you for your post. Could you tell me which pair it is? I will see if I can reproduce that.

[Edit] I think I have found out why the EA missed the trade now. There is a reason why the EA did not enter at the bar you pointed out. Because that's the 2nd bar to right of the support point for the TL. At that bar the support point haven't been validated yet (need 2 closed bar to the right) so the TL wasn't there.

In this chart, the EA should have entered the trade at the 3rd bar (the next bar to the bar you labeled). It didn't because I am still using the recent 3 bars to determine if there is a TL break. But since the opening price of the recent 3 bars are all below TL, so it decides that there was no TL break, hence no trade.

I will change to use the recent 4 bars to determine if there is a TL break, hopefully that will solve this problem and not to produce other problem.

[Edit] The bug fix is attached, please let me know if it works.


Thanks again!

aha

Quoting phoenix
hey AHA in testing your EA, i manually began going through the trades your EA made. i caught one trade that your EA missed (i think i caught it at least).

the trade should have been a sell on august 10, 2006, 4:00
at 1877.
Attached File(s)
File Type: mq4 Mouteki EA 0.1.mq4   18 KB | 1,375 downloads
Target: Trading naked.
  • #28
  • Edited 2:25am Sep 30, 2006 1:57am | Edited 2:25am
Merlin, thank you for the comment. .

BTW, can you give me the right to modify the post 5-6? I want to update the EAs with bug fixes.



Quoting merlin
my only comment is that this is a beautiful thread, a testament to the strength and kindness of our community here.
Target: Trading naked.
  • #29
  • Sep 30, 2006 3:08am Sep 30, 2006 3:08am
I really like this idea and post. The method that I have been trying to reproduce into an EA (unsucessfully using this EA creator tool that I found) is the Vegas 4 hour method. Have you tried coding it yet?
  • #30
  • Sep 30, 2006 4:39am Sep 30, 2006 4:39am
Hi Aha,

great work.

I have the same problem abt the AHA Bago indicator..

I cant get it to work, loads fine, but no arrows ? I have loaded a ton of indicators before, but this one i cant crack!?

Thanks

Quoting summitfx
great work- AHA THANKS when i go to load the ea on the chart- nothing happens- all my others work-any thoughts?
  • #31
  • Sep 30, 2006 6:17am Sep 30, 2006 6:17am
Hi AHA,

Thanks for starting this thread; I agree with Merlin's comments completely.

I am not up to speed with writing these things, however I have checked out the programming and I am impressed by your work.

I have a suggestion/request:

I prefer to do my testing in real time, however I work full time and I am not always able to check the charts at the 4 hour rollover times. I would like to be able to preset my entry conditions into a custom EA and have this monitor the OPEN price of the new 4HR bar and ENTER a trade if the conditions are met.

I see this as a simple EA that would execute a single entry order if the OPEN is above/below the trend line value, set the take profit and stop loss values and then do nothing else (i.e. turn itself off and wait for me to enter the next set of entry conditions and reset the EA).

This way I will be able to assess each entry, reject those that I do not like for some reason (low Reward/Risk ratio, trend line too steep, etc) and tweak the stop and targets to fit my individual version of the system.

My ideal input parameters are:
Trade Direction: LONG/SHORT
Parcel Size: xx contracts
Trend Line Value: (xx)x.xx(xx) [I want to enter this manually; I do not want the EA to calculate any values]
Stop Margin: (x)xx pips [I also want to enter this manually; I do not want the EA to calculate any values]
Take Profit Margin: (x)xx pips [I also want to enter this manually; I do not want the EA to calculate any values]
Entry Gap Filter: (x)xx pips [if the bar opens more than xx pips away from the Trend Line Value discard the trade]

I believe that many traders who work full time (or sleep at nights!) and have difficulty being at their computers at the exact 4HR rollover times would get great benefit from such an EA, whether they are using it in Demo or Live accounts for the Mouteki system or any other system that utilises 4HR bars.

I acknowledge that the request is off on a tangent to the original objective of your thread, however this thread is clearly where the EA writing skills reside; if you (or others) are able to assist I am grateful; if not I understand.

Regards,
Mike
  • #32
  • Sep 30, 2006 6:47am Sep 30, 2006 6:47am
aha,

I ran the Mouteki_EA on GBPUSD,H4 and noticed something unusual on the 2006.09.27 08:00 Order to Sell. Initially the TP is set to 1.8727 then 24 hours later the TP is changed to 1.8561. A substantial change. On the next bar the original TP would have been hit. The TP is changed three more times before the test closes. My understanding is that once the TP is set at the time the order is placed that it remain fixed. To see it change was unexpected so I bring it to your attention.

I have attached a screenshot of my test. Order 3 with the original TP is initiated on Step 6 (green). The changed TPs are in yellow. The test was run with default settings except for the lot size.

http://img137.imageshack.us/img137/3...ngtplr1.th.png

You have started an excellent thread. It takes a substantial amount of time, skill and effort to develop an EA. I appreciate your willingness to share your results.

hap
  • #33
  • Sep 30, 2006 7:16am Sep 30, 2006 7:16am
Quoting aha
Let me defence the TP rule for the order 605 first:

In the Mouteki system the TP is calculated based on the highest bar, not the bar will produce the largest TP. In the chart you posted, if we call the entry bar as bar 0, the EA actually followed the Mouteki system by calculating the TP using the bar 1, while you calculated the TP using bar 7.
Well, I understood that TP is calculated using the highest (lowest) bar that will be found between those two points of the TL. The other candles (outside those two points) are ignored.
The trend isyour friend untilis about to end...
  • #34
  • Sep 30, 2006 8:35am Sep 30, 2006 8:35am
Quoting aha
Here is the EA I made for Mouteki system. It is currently being tested by many members on this forum. You may downloaded here and either backtest it or forward test it.

You may test it silently, there is no obligation to post your findings (bug report, trade setup or results, ideas to modify, etc.) here. But if you would like to share your findings, please do it in an organized but convenient way.
Thanks for your PM re: MT4 testing inconsistencies. I've posted a chart below that shows one example of the problem. Two things are wrong here. First, the trade as shown is opened one bar later than it should be - the open of the preceding bar is also below the demand line. Secondly, there is a missing supply line (the dotted line marked A) that should cause the trade to be closed on a reversal and a new sell initiated. The Mouteki instructions state clearly that a support point must have 2 candles to the right and left that do not exceed the high/low of the support point so, even though the candle 2nd to the right has an equal high, the support point is still valid.

I'm not being critical of your EA - mine makes exactly the same mistake at the same place. I believe this may be due to a rounding error in the Tester. My EA is based on an indicator I wrote to draw the trend lines and the point in question is identified as a support point in the normal chart as well. Perhaps MT4 rounds differently in the regular chart than it does in the tester? We'll never know...

Thanks for your contributions and for taking the time to pursue this question.

Lou G
Old Benjamin was right
  • #35
  • Sep 30, 2006 8:43am Sep 30, 2006 8:43am
Sorry, file would not attach. I'll try again.
Old Benjamin was right
  • #36
  • Sep 30, 2006 8:50am Sep 30, 2006 8:50am
Whoops - I meant "a reversal and a SELL" at the cross of the dotted supply line. It's very early in the morning here...
Old Benjamin was right
  • #37
  • Sep 30, 2006 9:01am Sep 30, 2006 9:01am
Here's another example. This is a live chart (rather than from the Tester). The dotted demand line does not appear when the EA runs, thus missing a good short trade. There is a double bottom at point A which, per Mouteki rules, should be a support point.
Old Benjamin was right
  • #38
  • Sep 30, 2006 9:03am Sep 30, 2006 9:03am
Better picture.
Old Benjamin was right
  • #39
  • Sep 30, 2006 9:06am Sep 30, 2006 9:06am
The more I look at this, the more I think that the regular chart is rounding to 4 decimal places and the Strategy Tester is not.
Old Benjamin was right
  • #40
  • Sep 30, 2006 9:50am Sep 30, 2006 9:50am
Yeah, I did code the original 4 hr vegas tunnel. But the result was pretty bad because the lagging effect of the weekly trend detection method. The 4H trend may have changed the trend for days but the system still produce signals according to old trends.

They are using something new now but I haven't coded that yet.


Quoting altapowder
I really like this idea and post. The method that I have been trying to reproduce into an EA (unsucessfully using this EA creator tool that I found) is the Vegas 4 hour method. Have you tried coding it yet?
Target: Trading naked.
3 traders viewing now