EA Factory | Page 6 | 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

  • #101
  • Oct 1, 2006 9:53pm Oct 1, 2006 9:53pm
Would you please explain the advantage of this system against Sidus or Bago, if you are familar with one of those two system?

aha


Quoting stricko108
hi aha


I'm wondering if you have time if you could make a EA for the following system.

when 10 and 40 ema cross up it is a buy signal confirmed
by the body of the candle ABOVE the ema's, opposite for
shorts, 30 min chart. You wait for bar close as well.
No stops, your always in the market, reversing position when you have a cross in the opposite direction and a candle body that supports the cross up or down that is NOT touching the EMA's.
You open with 2 positions, close one at 40 pips, let the
2nd one ride till you get valid signal to reverse position.
Target: Trading naked.
  • #102
  • Oct 2, 2006 12:59am Oct 2, 2006 12:59am
hello

regarding BAGO_EA, will it open position itself?

when i set up on USDCHF, the upmove arrow did not show up eventhough the swissy went up more than 200+pips.
{Link Removed}]
  • #103
  • Oct 2, 2006 1:09am Oct 2, 2006 1:09am
Quoting aha
About Mouteki EA, here is something I posted on another thread. If you are interested in contributing your computer time, please let me know.

And here is the link if you would like to know the context:

http://www.forexfactory.com/forexfor...d=1#post109699
Hi aha.

If you have the time, would you be able to implement an EA using Mouteki's trading method with two modifications?

1) we enter a trade with 3 lots instead of one
2) when a trade moves in the desired direction that we want we take one lot off at 33% for a partial profit and another lot off at 66% for another partial profit.

Everything else remains the same.

I truly think with simple and effective money management we can significantly improve the performance of mouteki's trading method.

if you (or anybody can do this) it would be greatly, greatly appreciated.

phoenix

p.s. i've been trying to code a partial profit EA for the last day and a half (literally) and have been killing myself doing it.
  • #104
  • Oct 2, 2006 1:38am Oct 2, 2006 1:38am
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.
Hey aha.

may i suggest another rule for consolidation?

1) large upper shadow on candles (say greater than 50% of the body of the candle) after an steep upswing or large lower shadows after large down swing. this signifies selling pressure and buying pressure respectively.

2) successively smaller bodies as more recent candles appear. each successive candle is 90% or less of the previous candle. this shows consolidation.

i don't know if this will make help in determining consolidation areas mechanically or whether it's not specific enough...

Great Job AHA!

phoenix

p.s. way to go through the my "false" bugs. i'll do a better job next time. i'm interested in looking at mouteki v0.2
  • #105
  • Oct 2, 2006 1:57am Oct 2, 2006 1:57am
hi aha et al.

i've noticed that none of mouteki's trendlines extend back more than 25-30 candles (in his GBP/USD july analysis slides). that amounts to less than one week.

he also mentions in his documentation that he places a premium on the most recent price action as it contains the most recent information. In addition he mentions he redraws all his trendlines at the begining of the week.


could this potentially be one of mouteki's filters?

he just doesn't trade any trendline that extends greater than 1.5 weeks away into the past?

and he waits for shorter term horizon trendlines to form and only trades on that?

I'll take a deeper look into this. if it looks promising i'll let you know.

Perhaps we can code this neglecting of trendlines greater than 1.5 weeks, into the mouteki ea and have it as a feature you turn on or off.

phoenix
  • #106
  • Oct 2, 2006 2:23am Oct 2, 2006 2:23am
well I guess if you take all bago signals based on cross and rsi only, it doesnt' work out to well, but this system it seems to handle well on the basic rules, and I find it interesting that your never out of the market with it. Apparetnly on pound, it produces about 800 pips per month following the basic rules.

No worries if your to busy, just thought it might be a good one.


Quoting aha
Would you please explain the advantage of this system against Sidus or Bago, if you are familar with one of those two system?

aha
  • #107
  • Oct 2, 2006 2:28am Oct 2, 2006 2:28am
Hi Aha and Everybody!

I'm posting an update to my Flyer! EA. This updated version will allow the EA to recover in case of power failure, or if you want to shut the program down, or even simply change the chart. This prevents it from "forgetting" previous orders.

Also, you can run the EA live on any chart timeframe you want. But if you backtest it, Strategy Tester must be set to use the M1 setting. Otherwise erroneous results will occur.

GirlFlyer
Attached File(s)
File Type: mq4 Flyer!.mq4   16 KB | 898 downloads
  • #108
  • Oct 2, 2006 2:36am Oct 2, 2006 2:36am
Quoting phoenix
Hi aha.

If you have the time, would you be able to implement an EA using Mouteki's trading method with two modifications?

1) we enter a trade with 3 lots instead of one
2) when a trade moves in the desired direction that we want we take one lot off at 33% for a partial profit and another lot off at 66% for another partial profit.
I did manual backtest for this, but it is basically scaling out. I know it seems good at first time, but there are long trends you really need to take with full power to cover up the losses and make some extra profit.

But consider it just MHO

bluefish
  • #109
  • Oct 2, 2006 3:09am Oct 2, 2006 3:09am
Quoting GirlFlyer
Hi Aha and Everybody!

I'm posting an update to my Flyer! EA. This updated version will allow the EA to recover in case of power failure, or if you want to shut the program down, or even simply change the chart. This prevents it from "forgetting" previous orders.

Also, you can run the EA live on any chart timeframe you want. But if you backtest it, Strategy Tester must be set to use the M1 setting. Otherwise erroneous results will occur.

GirlFlyer
i run strategy tester but the result is poor... is this being forward tested??
{Link Removed}]
  • #110
  • Oct 2, 2006 4:11am Oct 2, 2006 4:11am
Quoting kenari05
i run strategy tester but the result is poor... is this being forward tested??
I am forward testing it... so far results look ok. But let me know the settings you used on strategy tester and what currency pair... might be your settings.
  • #111
  • Oct 2, 2006 7:05am Oct 2, 2006 7:05am
Quoting GirlFlyer
I am forward testing it... so far results look ok. But let me know the settings you used on strategy tester and what currency pair... might be your settings.
I noticed you had a Modelling Quality of only 25%. I made the experience that the MT4 strategy tester is very unreliable as long as you had not a minimum of about 90% Modelling Quality. I saw dozens of EA's that would make you a millionaire in only one year but in forward tests they all were losers.

Anyone who makes a backtest should make it with the minimum requirements (90% Modelling Quality). Anything below is like guessing.
  • #112
  • Oct 2, 2006 8:02am Oct 2, 2006 8:02am
Quoting katzenhai2
I noticed you had a Modelling Quality of only 25%. I made the experience that the MT4 strategy tester is very unreliable as long as you had not a minimum of about 90% Modelling Quality. I saw dozens of EA's that would make you a millionaire in only one year but in forward tests they all were losers.

Anyone who makes a backtest should make it with the minimum requirements (90% Modelling Quality). Anything below is like guessing.
I have tried resolving this problem. I have run additional tests making sure that I have all historical data on all timeframes for the period run on strategy tester and still get a 25% model quality. If anybody can help to solve why this is so, I would be up for suggestions, because I'm at wits end on how to improve it.

GirlFlyer
  • #113
  • Oct 2, 2006 8:30am Oct 2, 2006 8:30am
Ok, I am getting quite a few responses about the 25% modelling quality of my tests which make people doubt the veracity of the EA which I can understand. However, further research shows that on the M1 chart which I am running the strategy tester it is impossible to get better than 25% according to the method strategy tester uses to calculate that percentage. Quoted from the Metatrader MQ4 website (see link at: http://articles.mql4.com/93 )

Quoted from above link:

"Reasoning from the same formula, the one-minute data modelled cannot be of quality exceeding 25 per cent since data of a smaller period are not used for modelling of these data. For client terminal, there are simply no data of a smaller period, but there are tick data. However, tick data are not stored in the client terminal by standard means.

In themselves, one-minute data are the most detailed data about the price movement. And, using them, one can model the price movement inside of bars of larger periods. At that, discrepancy between the real price movement and the ticks modelled will not be large due to presence of one-minute reference points: Open, High, Low, Close. These reference points form the first approximation of the price movement function (the first-order approximation). The higher is the value of the period modelled, the more there are reference points and the more exact is the modelling. And vice-versa, the lower the value of the modelled period is, the lower is the modelling quality. For example, up to 240 one-minute reference points can be used to model a one-hour bar. When modelling a 5-minute bar, the amount of reference points drops to 20.

As we mentioned before, there are just 4 reference points for a one-minute bar to be formed! This is why the rating of modelling quality for one-minute bars does not exceed 25 per cent. However, if one uses signals received from data of a larger timeframe when testing on one-minute timeframe, the data modelled should, of course, be rated much higher because the price movement within the bar of timeframe larger than the on-minute one is considered."

End of Quote

So, if you read this and understand it, it is now shown that better than 25% on the M1 is impossible.

Thus, the EA test results are probably about as accurate as they can get.

I hope this helps for people to understand that my results are not bogus.

GirlFlyer
  • #114
  • Oct 2, 2006 8:32am Oct 2, 2006 8:32am
Hehe, other's code is always hard to read, same to yours, even with comments.
Quoting aha
Man, your code is hard to read. . I didn't find a single line of comment.

Could you briefly discribe your system a little before I dive in?

aha
  • #115
  • Oct 2, 2006 8:37am Oct 2, 2006 8:37am
Quote
So, if you read this and understand it, it is now shown that better than 25% on the M1 is impossible.
Well, then ok. I never used the M1 for testing.
  • #116
  • Oct 2, 2006 9:22am Oct 2, 2006 9:22am
Quoting GirlFlyer
I have tried resolving this problem. I have run additional tests making sure that I have all historical data on all timeframes for the period run on strategy tester and still get a 25% model quality. If anybody can help to solve why this is so, I would be up for suggestions, because I'm at wits end on how to improve it.

GirlFlyer
please see my posts in this thread for explanation about the modelling quality

cheers
RoBiK
I'm not insane; my mother had me tested.
  • #117
  • Oct 2, 2006 9:54am Oct 2, 2006 9:54am
Ken,

Welcome here!

For you question: No EA is supposed to catch every move.... and this EA miss a lot of them, . But the important is, in the end, this seems to be a winner.

Anyone else who have data older than 2004, please test the Bago_EA and post the result. I only have the data since then.

aha


Quoting kenari05
hello

regarding BAGO_EA, will it open position itself?

when i set up on USDCHF, the upmove arrow did not show up eventhough the swissy went up more than 200+pips.
Target: Trading naked.
  • #118
  • Oct 2, 2006 10:00am Oct 2, 2006 10:00am
This sounds like a good exiting strategy. I will implement it when I find time. And please remind me if I forget. If someone is interested to implement this, please feel free to do it and I can review the code if you have commented it sufficiently (just joking with you, GP2X, )

aha


Quoting phoenix
Hi aha.

If you have the time, would you be able to implement an EA using Mouteki's trading method with two modifications?

1) we enter a trade with 3 lots instead of one
2) when a trade moves in the desired direction that we want we take one lot off at 33% for a partial profit and another lot off at 66% for another partial profit.

Everything else remains the same.

I truly think with simple and effective money management we can significantly improve the performance of mouteki's trading method.

if you (or anybody can do this) it would be greatly, greatly appreciated.

phoenix

p.s. i've been trying to code a partial profit EA for the last day and a half (literally) and have been killing myself doing it.
Target: Trading naked.
  • #119
  • Oct 2, 2006 10:04am Oct 2, 2006 10:04am
Actually I disagree with Mouteki on this one. Long term trendline can be very strong and once break it will result in big moves. But of course not every long term trend line will have that magic effect, just like short term trend lines.

You guys can read what Raczekfx posted in Bago thread to see the power of daily even weekly trendline.


Quoting phoenix
hi aha et al.

i've noticed that none of mouteki's trendlines extend back more than 25-30 candles (in his GBP/USD july analysis slides). that amounts to less than one week.

he also mentions in his documentation that he places a premium on the most recent price action as it contains the most recent information. In addition he mentions he redraws all his trendlines at the begining of the week.


could this potentially be one of mouteki's filters?

he just doesn't trade any trendline that extends greater than 1.5 weeks away into the past?

and he waits for shorter term horizon trendlines to form and only trades on that?

I'll take a deeper look into this. if it looks promising i'll let you know.

Perhaps we can code this neglecting of trendlines greater than 1.5 weeks, into the mouteki ea and have it as a feature you turn on or off.

phoenix
Target: Trading naked.
  • #120
  • Oct 2, 2006 10:08am Oct 2, 2006 10:08am
I see what you mean, Bluefish. Thanks for sharing your findings here. Just one question. When you see this exiting strategy exit too early in long term trends, did you see the long term trends usually provide reenter opportunity?



Quoting bluefish
I did manual backtest for this, but it is basically scaling out. I know it seems good at first time, but there are long trends you really need to take with full power to cover up the losses and make some extra profit.

But consider it just MHO

bluefish
Target: Trading naked.
0 traders viewing now