EA Factory | Page 4 | 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

  • #61
  • Oct 1, 2006 12:12am Oct 1, 2006 12:12am
Fozzy method is another method that is doing great recently. Stockwet and Tradeigel had made an EA for this method and my test is their EA (after bug fix) follows the system pretty well. Here is their thread:

http://www.forexfactory.com/forexfor...ead.php?t=8527

The attached is my backtest result of Fozzy EA for EURUSD in 2006. It did not survive the first 5 monthes.

If you backtest it only in Sept and Aug, the performance was pretty good, which is consistent with what we saw on Fozzy's thread.
Attached File(s)
File Type: zip Fozzy 0.1.zip   7 KB | 841 downloads
Target: Trading naked.
  • #62
  • Oct 1, 2006 12:17am Oct 1, 2006 12:17am
Anyone knows a reliable EA available for Kolachi?
Target: Trading naked.
  • #63
  • Oct 1, 2006 3:31am Oct 1, 2006 3:31am
Quoting aha
Hi Divinefund,

Welcome to be here. I know you follow Mouteki thread a lot and your opinion is valued here.

I am actually with you on the first point. If you will inspect my EA you will find that the EA does exactly what you said.

I don't quite understand what you said about the PR part. Could you be more accurate?

aha
Sorry for too brief on the last paragraph, what I mean is mounteki did mention that he will not use the highest point as projected project when there is a reversal trend, hence he use the next previous higher point as his projected profit. If you have he PPT for GBPUSD of part 2 slide #4.

Another issue I did say before that when both trend cross and the price cross both trends ( I use zone C as my concept ), mouteki did not say what to do but I rather do nothing until the price fall only into one of the trend. I guess your EA also does nothing and it shouldn't close all the trades as it is not the trend reversal case.
  • #64
  • Oct 1, 2006 3:39am Oct 1, 2006 3:39am
Sorry for the slightly tangential interlude, but does anyone remember the series of links containing tutorials for EA programming? I'm trying to learn writing/compiling EA's for myself and I just know the links were contained in these forums somewhere, but forgot where. I'd appreciate any help.
  • #65
  • Oct 1, 2006 4:57am Oct 1, 2006 4:57am
http://www.forexfactory.com/forexfor...ead.php?t=7637

here you go.

Quoting weghio
Sorry for the slightly tangential interlude, but does anyone remember the series of links containing tutorials for EA programming? I'm trying to learn writing/compiling EA's for myself and I just know the links were contained in these forums somewhere, but forgot where. I'd appreciate any help.
Target: Trading naked.
  • #66
  • Edited 5:36am Oct 1, 2006 5:04am | Edited 5:36am
Just checked, you are right on this one. [Edit] Check again, what 's Mouteki says was "Area of consolidation before market reversal, none of these highs are valid in my perspective for price projection". so it's not the trend reversal make us ignore those highs after the first point, it's the consolidation. I find it difficult to define "consolidation", but I will use the following rules: on a trend reversal, when the PP calculated based on the highs after the first point is smaller (consolidating) than that based on the highs between the first point and the 2nd point, use latter PP for calculating ST and TP.

About your zone c strategy, yes, it's probably better not to close the trade immediately but wait for further signs of reversal.


Quoting divinefund
Sorry for too brief on the last paragraph, what I mean is mounteki did mention that he will not use the highest point as projected project when there is a reversal trend, hence he use the next previous higher point as his projected profit. If you have he PPT for GBPUSD of part 2 slide #4.

Another issue I did say before that when both trend cross and the price cross both trends ( I use zone C as my concept ), mouteki did not say what to do but I rather do nothing until the price fall only into one of the trend. I guess your EA also does nothing and it shouldn't close all the trades as it is not the trend reversal case.
Target: Trading naked.
  • #67
  • Oct 1, 2006 5:17am Oct 1, 2006 5:17am
Quoting accrete
Greetings all. I'll share my 2pips worth on auto-trendline creation tools from past live/forward testing. There are a couple very nice indicators out there that do a decent job of auto creating the TLs, though my forward testing over a period that included about 90days showed me that human-created TLs are much more accurate. Sometimes the Robot in the Machine just went buggy and you would look at your chart and wonder where it got the info to draw a particular TL.

This is why i believe that Mikes post is valuable and worth a second look (a copy paste is below my comments here).

On EAs in general? I'm relatively new to EA forward testing (only 5 weeks in demo w/a handful of EAs running 24/6). I have yet to see an EA that was a set and forget +pip machine. If one were to implement an EA that by its very design needed some initial setup, such as what i read in Mikes suggestion, I think at least a couple things would take place; 1) The trader would continue to build his/her technical analysis skills while studying the charts, 2) become proficient at drawing accurate trend lines. Both of these are good things in themselves and would build skills the trader can use anytime, anywhere.

....then the icing on the cake...Once the user sets these trend lines, places a "trade in Long" (or short) only situation if called for (maybe the trend is so big it just makes sense to trade a penant breakout in the direction of general trend for a given situation?) then let the EA do it's thing and get you into the trade while you have a life?

I will look around and see what i can find in code to assist.

If this is a "Only this or that" EA thread, pardon my comments. Just trying to think outside the EAbox. I've been researching how to manage EAs for a while now and believe that getting the Human-Heart into the Robotic-Mind is the way to pips.

Cheers,
Thom
Hi Thom,

Thanks for the comments, for furthering my thought process and putting the case so elegantly.

However, I note AHA's comments in his second post:

"But again, there is no room in this thread to argue about the integrity of backtesting method."

and I have no desire to hijack this thread. I believe that we should continue our discussion and the evolution of an EA based trading tool in a separate thread; possibly under the Metatrader Forum section?

I see it as a very simple few lines of code; I simply do not have the programming experience to get it down and working; my attempts at MT4 code writing to date have been feeble and frustrating to say the least.

I will open a new topic with links to my initial post here and your response.

Regards,
Mike
  • #68
  • Oct 1, 2006 5:18am Oct 1, 2006 5:18am
Quoting aha
Hi Thom, Mike,

As you can see from my other posts this is not a "only this or that thread". I think it's a great idea to define the parameters of TL trading, make an script to do the boring job so human don't have stay in front of the computer, and still leave the major decisions to human. What You and Mike described is more like a trading tool, not an EA. It should be easy to build but can't be tested on the strategy tester so can't provide statistics to help you understand the win/loss ratio or max drawdown or such such of your TL trading strategy, or to optimize the parameters against the historical data.

I am getting my hands full right now. Are there anyone volunteering for this one? I can help if you need guidance.

aha
Hi Aha,

Thanks for your response. I agree that I am looking for an EA based trading tool that will provide a conditional "execute only" function at specific times when I am unable to be at the computer.

No, it is not able to be backtested as a true EA is; however, such a tool will assist me greatly in forward testing any system. It will take a little longer, but the real time results I obtain will provide me with the statistics I need to build confidence in any system.

As noted in my response to Thom, I have no desire to hijack this thread which is really about true EA development.

I will open a new thread under the Metatrader section; if any members are able to supply a simple piece of code to assist I would be most grateful.

Thanks,
Mike
  • #69
  • Oct 1, 2006 5:54am Oct 1, 2006 5:54am
Quoting aha
Just checked, you are right on this one. [Edit] Check again, what 's Mouteki says was "Area of consolidation before market reversal, none of these highs are valid in my perspective for price projection". so it's not the trend reversal make us ignore those highs after the first point, it's the consolidation. I find it difficult to define "consolidation", but I will use the following rules: on a trend reversal, when the PP calculated based on the highs after the first point is smaller (consolidating) than that based on the highs between the first point and the 2nd point, use latter PP for calculating ST and TP.

About your zone c strategy, yes, it's probably better not to close the trade immediately but wait for further signs of reversal.
Thanks for the reply. It is difficult to define the trend reversal. As my understanding is when price say in the down trend below the demand trend and then suddenly or gradually move above the supply trend. But this scenario is very unlikely to happen as trend line change whenever their point A and B change. Hence I just wonder how do you put the idea into the program as you may need to keep the previous trend lines for reference or recompute it.
  • #70
  • Oct 1, 2006 6:14am Oct 1, 2006 6:14am
Hi, thought I would add a post in here also for my EA. I've seen many of the EA's used here on FF and many are pretty complex such as Firebird. Firebird is a good EA and seems to work very well, but I also wanted to test one of my own. I call it FLYER! It is a moving average EA based on the principles of the "Sidus" method with a few minor differences and some fine tuning. Preliminary results on backtests were good.

Try it out and post some results or ask questions. PM me if you wish as well.

See my thread in the main forum for further discussion.

GirlFlyer
Attached File(s)
File Type: zip FLYER!.zip   153 KB | 1,088 downloads
  • #71
  • Oct 1, 2006 6:14am Oct 1, 2006 6:14am
Thanks for hans_gregor and divinefund's findings and suggestions, here we got the Mouteki v0.2 released.

Here are the modifications:

// Ver 0.2
// Modiciations:
// 1. On a trend reversal, when the PP calculated based on the high
// after the first point is smaller (consolidating) than that based
// on the high between the first point and the 2nd point, use
// latter PP for calculating ST and TP.
// 2. When both TL are broken, only signal this special situation, not
// to close any trades. When price break the Top TL again from that
// situation, close all the short trades and let the long trades
// run. When price break the bottom TL again from that situation,
// close all the long trades and let the short trades run.
// Ver 0.1
// Modifications:
// 1. Allow EA to enter a trade even 3 bars after TL breaks. This is
// because sometimes we got whipsaw back below TL and then breaks
// again, or sometimes, the break is not detectation by only
// checking the recent two bars.

Thanks again and you are welcomed to test it out.
Attached File(s)
File Type: mq4 Mouteki EA 0.2.mq4   19 KB | 1,111 downloads
Target: Trading naked.
  • #72
  • Oct 1, 2006 6:23am Oct 1, 2006 6:23am
This is how I take it: When price in zone C, price>TTL (top TL) and price<BTL (bottom TL). If a TL change and break this relationship, then price is not in zone z anymore, we will use the normal TL break rules to deal with it. If no TL changes, and price move to break BTL again from zone c, enter long and close short, break TTL again from zone c, enter short and close long.

Let me know if there is a problem here.


Quoting divinefund
Thanks for the reply. It is difficult to define the trend reversal. As my understanding is when price say in the down trend below the demand trend and then suddenly or gradually move above the supply trend. But this scenario is very unlikely to happen as trend line change whenever their point A and B change. Hence I just wonder how do you put the idea into the program as you may need to keep the previous trend lines for reference or recompute it.
Target: Trading naked.
  • #73
  • Oct 1, 2006 6:28am Oct 1, 2006 6:28am
Welcome to this thread, GirlFlyer! Thanks for sharing your EA. I will test it in a minute. It's much better working together than working alone.



Quoting GirlFlyer
Hi, thought I would add a post in here also for my EA. I've seen many of the EA's used here on FF and many are pretty complex such as Firebird. Firebird is a good EA and seems to work very well, but I also wanted to test one of my own. I call it FLYER! It is a moving average EA based on the principles of the "Sidus" method with a few minor differences and some fine tuning. Preliminary results on backtests were good.

Try it out and post some results or ask questions. PM me if you wish as well.

See my thread in the main forum for further discussion.

GirlFlyer
Target: Trading naked.
  • #74
  • Oct 1, 2006 6:36am Oct 1, 2006 6:36am
Wow! your test result blowed me away. Perfect profit curve, high rate of return, high win/loss ratio, lots of signals, is this the holy grail!?

Do you mind to share your system here? Like the entry/exit condition, filters? Because I made an EA for Sidus before and I know his system can't achieve this kind of result. [Edit] Ok I see your system in the code. I will backtest it now. Your modelling quality was a little low though.

aha


Quoting GirlFlyer
Hi, thought I would add a post in here also for my EA. I've seen many of the EA's used here on FF and many are pretty complex such as Firebird. Firebird is a good EA and seems to work very well, but I also wanted to test one of my own. I call it FLYER! It is a moving average EA based on the principles of the "Sidus" method with a few minor differences and some fine tuning. Preliminary results on backtests were good.

Try it out and post some results or ask questions. PM me if you wish as well.

See my thread in the main forum for further discussion.

GirlFlyer
Target: Trading naked.
  • #75
  • Oct 1, 2006 6:57am Oct 1, 2006 6:57am
Aha,


I'm more than happy to share ideas and information about the system. Yes, the results are astounding. I would like a little better modeling quality, too as you say. I'm not quite sure what to do to my dataset to improve it yet. Still working on that angle. But other than that, ask away, test away!

GirlFlyer
  • #76
  • Oct 1, 2006 7:02am Oct 1, 2006 7:02am
I have finished reading your code. You seemed using a filter that the sidus entry condition has to be valid for at least 5 bars to avoid whipsaw, does that make this much difference? That was clever and I saw another coder applied something similar to Bago system.

Also, I don't quite understand why this EA has to be run on 1M timeframe. Your code seems to be independent to whatever timeframe it runs. Am I missing something here?


Quoting GirlFlyer
Aha,


I'm more than happy to share ideas and information about the system. Yes, the results are astounding. I would like a little better modeling quality, too as you say. I'm not quite sure what to do to my dataset to improve it yet. Still working on that angle. But other than that, ask away, test away!

GirlFlyer
Target: Trading naked.
  • #77
  • Oct 1, 2006 7:15am Oct 1, 2006 7:15am
Quoting aha
I have finished reading your code. You seemed using a filter that the sidus entry condition has to be valid for at least 5 bars to avoid whipsaw, does that make this much difference? That was clever and I saw another coder applied something similar to Bago system.

Also, I don't quite understand why this EA has to be run on 1M timeframe. Your code seems to be independent to whatever timeframe it runs. Am I missing something here?
Aha, I have not quite understood all the mechanics of MT4 yet, but the reason why I have it run on the 1minute chart is it seems that especially in the strategy tester, it provides for a more accurate entry point. The bid and ask values inherent to the trading system have to be accurate to give good results and the M1 chart gives the best results for strategy tester because if you run it on a M30 chart during strategy tester, it has to guess what the bid and ask values are doing that 30 minute session. If during live testing, the EA is run on a tick by tick basis as data comes in, theoretically it should be fine to run it on any time frame. But I've run into other EA's from other programs (like VTTrader) that seem to wait until the current period is completely over before making a decision to trade or not. That cannot happen here, it must be able to make a decision to trade or not within a very fast time frame because the MA's can cross very rapidly especially during one 30 min session when a news break occurs. That's my reasoning behind the M1 chart.
  • #78
  • Oct 1, 2006 7:25am Oct 1, 2006 7:25am
Quoting aha
I have finished reading your code. You seemed using a filter that the sidus entry condition has to be valid for at least 5 bars to avoid whipsaw, does that make this much difference? That was clever and I saw another coder applied something similar to Bago system.

Also, I don't quite understand why this EA has to be run on 1M timeframe. Your code seems to be independent to whatever timeframe it runs. Am I missing something here?
On the 5 period check... that coding was there primarily to make sure the entry was made at the correct moment and that it was definitely trending in the right direction, otherwise the program would and could jump in a trade at any point after the crossing and yes, they do jump around in there often at those points. Once the WMA's cross the EMA's... the EA makes its decision and sticks with it even if the WMA's reverse and go back up. The S/L then takes over the exit if a reversal occurs. This way when whipsawing is going on for an extended period of time within a range, multiple exits and entries are not made. It will only produce 1 trade until the range is broken. After the range is broken, either up or down, it will either take a profit or get stopped out, and then wait for another entry. This is what allows this EA to keep from doing multiple entries/exits over a very short period of time thus losing money due to whipsaw conditions.

Hope this helps,
GirlFlyer
  • #79
  • Oct 1, 2006 7:33am Oct 1, 2006 7:33am
I thought if we run backtest on ticktotick bases it will go downto the least timeframe available so it won't make any difference. But I have never tested MT4 on this yet.

I suppose we can run a quick test for just one or two days of data on both low and high timeframe and checked the entry price to find it out. But my computer is tied up in the backtesting of your EA now.... can anyone do that test?

Ok, if you don't have 1M data on your computer it may make big difference because low timeframe data will be unavailable for high time frame test.

Here is a good post that may help you improve your back test quality and many other useful staff(if you haven't read it yet):

http://www.forexfactory.com/forexfor...ead.php?t=7637



Quoting GirlFlyer
Aha, I have not quite understood all the mechanics of MT4 yet, but the reason why I have it run on the 1minute chart is it seems that especially in the strategy tester, it provides for a more accurate entry point. The bid and ask values inherent to the trading system have to be accurate to give good results and the M1 chart gives the best results for strategy tester because if you run it on a M30 chart during strategy tester, it has to guess what the bid and ask values are doing that 30 minute session. If during live testing, the EA is run on a tick by tick basis as data comes in, theoretically it should be fine to run it on any time frame. But I've run into other EA's from other programs (like VTTrader) that seem to wait until the current period is completely over before making a decision to trade or not. That cannot happen here, it must be able to make a decision to trade or not within a very fast time frame because the MA's can cross very rapidly especially during one 30 min session when a news break occurs. That's my reasoning behind the M1 chart.
Target: Trading naked.
  • #80
  • Oct 1, 2006 7:38am Oct 1, 2006 7:38am
Quoting aha
I thought if we run backtest on ticktotick bases it will go downto the least timeframe available so it won't make any difference. But I have never tested MT4 on this yet.

I suppose we can run a quick test for just one or two days of data on both low and high timeframe and checked the entry price to find it out. But my computer is tied up in the backtesting of your EA now.... can anyone do that test?

Ok, if you don't have 1M data on your computer it may make big difference because low timeframe data will be unavailable for high time frame test.

Here is a good post that may help you improve your back test quality and many other useful staff(if you haven't read it yet):

http://www.forexfactory.com/forexfor...ead.php?t=7637
Aha,

I have 2 years of M1 data and M30 data and 10 years of D1 data in my system. So, the EA ran with those data sets. If I ran the EA in strategy tester on any setting besides M1, the results were downright crappy.
0 traders viewing now