the impossible EA? | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Trading forex profitable is impossible! 269 replies

It's impossible to lose money in Forex!!! 91 replies

long-run profits impossible with short-term trading? 39 replies

Real Scalping, Impossible? 15 replies

Attachments: the impossible EA?
Exit Attachments

the impossible EA?

  • #1
  • First Post: Edited 5:20pm Feb 26, 2009 2:26pm | Edited 5:20pm
I've already consulted with Ronald Raygun about this idea. He's working with me on it, and he suggested I present my idea here as well so we can get some more opinions.

Please make allowances for the fact that I know nothing at all about programming, so I can not 'speak your language', as it were. But I do hope you will find this idea interesting nonetheless:

The purpose of the EA is to identify tick movement (#1) within the parameters of the following two filters:

Think of these 3 frames of reference:
1. Ticks (either up or down)
within:
2. Time (in seconds)
further filtered by:
3. Divergence (% of Bulls vs. Bears)

For example, 100 consecutive ticks (either up or down), within the last 250 seconds, 65% divergence factor (either direction).

So, with those settings I would have been notified when there had been a trending of 65 of the last 100 ticks (65/35), regardless of trend direction, so long as it had all happened within the past 250 seconds.

All 3 parameters need to be left blank and fully customizable.

That's the meat of it. Additional features could be various types of alerts, money management, on/off session times, etc.

The original idea was only to make a sort or indicator, but if this can be turned into a fully operational EA that takes trades, so much the better!

I would love to get your feedback on this idea!
" Average traders look to the left, great traders look to the right."
  • #2
  • Feb 26, 2009 11:04pm Feb 26, 2009 11:04pm
This is definitely something which will require dynamic arrays of sorts to hold the tick, direction, and time data. I was thinking 3 parallel arrays of sorts:

TickValue[]
TickDirection[]
TickTime[]

Each tick is assigned a value, direction, and time. The problem I'm having is working with arrays in MT4. How do I adjust the size of the array without losing the internal data?
  • #3
  • Feb 27, 2009 12:24am Feb 27, 2009 12:24am
you will only need one array instead of keeping a direction just store each tick as either a 1 or a -1. Time isn't needed and can be kept track of with a variable.

Is this going to be a rolling 250 seconds or will everything get reset every 250 seconds?
  • #4
  • Feb 27, 2009 12:53am Feb 27, 2009 12:53am
As I understand it, it's a rolling period.
  • #5
  • Edited 7:46am Feb 27, 2009 7:06am | Edited 7:46am
Thanks alot guys for coming in!

Each of the 3 variables (tick, time, divergence) are adjustable, from "1 to 99999." As for the momentary status of these variables, they will only look back as far as the user sets them up as. I think that I will be using it primarily for very short time frames (as in my example), but with the proper programming there's no reason it couldn't also be used with much larger filter numbers equally well.

So using the same numbers as before, the EA would have been looking back 100 ticks and/or 250 minutes (which ever comes first), the whole time looking for 65 out of those 100 ticks that went the same direction (either Buy or Sell) during that very limited period. That period being determined by the "and/or" variables of the first 2 filters.

Meaning, once either 101 ticks and/or 251 minutes has elapsed, the EA "forgets" the data of the tick and/or the minute furthest away and replaces it with the data from the newest one -- for the purpose of finding those 65 ticks that have moved in the same direction, but not necessarily consecutively.

Hope that helps. You guys are great! Couldn't do this without you.

Chance
" Average traders look to the left, great traders look to the right."
  • #6
  • Edited 9:59pm Feb 27, 2009 9:57am | Edited 9:59pm
...deleted...
  • #7
  • Feb 27, 2009 10:37am Feb 27, 2009 10:37am
Quoting magicchance
Thanks alot guys for coming in!

Each of the 3 variables (tick, time, divergence) are adjustable, from "1 to 99999." As for the momentary status of these variables, they will only look back as far as the user sets them up as. I think that I will be using it primarily for very short time frames (as in my example), but with the proper programming there's no reason it couldn't also be used with much larger filter numbers equally well.

So using the same numbers as before, the EA would have been looking back 100 ticks and/or 250 minutes (which...
If my understanding correct:

You want the ticks counts till 100.
within that 100, you want to count which 65 ticks come first(UP or DOWN).

Am I correct?

If within that 100 ticks, no 65 ticks detected, reset to 0 back?

Then how the trade entry should be when let say 65 ticks within 100 ticks is UP, is it time to BUY?(Entry)...when is Exit?
  • #8
  • Feb 27, 2009 10:37am Feb 27, 2009 10:37am
Quoting Ronald Raygun
This is definitely something which will require dynamic arrays of sorts to hold the tick, direction, and time data. I was thinking 3 parallel arrays of sorts:

TickValue[]
TickDirection[]
TickTime[]

Each tick is assigned a value, direction, and time. The problem I'm having is working with arrays in MT4. How do I adjust the size of the array without losing the internal data?
perhaps a swap buffers should help retaining the internal data
-1128 Pips since 6 August 2007
  • #9
  • Edited 11:26am Feb 27, 2009 11:08am | Edited 11:26am
Hi Akukaya, thanks for helping out here!

"You want the ticks counts till 100. Within that 100, you want to count which 65 ticks come first(UP or DOWN)."

Not only 100. Maybe 100, or 20, or 741 -- what ever they are set at by the user. The same applies to Time and Divergence. All 3 settings need to be totally open and independant of the other two. It is also important to remember that the EA is not looking for 65 consecutive ticks in any one direction.

"If within that 100 ticks, no 65 ticks detected, reset to 0 back?"

No. Collection of data begins where you are and goes back in time according to the setting of Tick & Time. Once those limits are exceeded, the oldest data is "forgotten", but all the newer data remains. The ONLY data "zeroed-out" is the part falling off the old end of the data stream as it is being replaced by the new.

"Then how the trade entry should be when let say 65 ticks within 100 ticks is UP, is it time to BUY?(Entry)...when is Exit?"

The simple answer is "yes." But it probably isn't enough information for a stand alone EA. I happen to like the RSI very much. Over here in this thread is a very good RSI EA being discussed, and I can see the 2 being combined. That to me would create a very good stand alone EA model: http://www.forexfactory.com/showthread.php?t=154687

Thanks again for your help, and be sure to let me know if you have any more questions.
" Average traders look to the left, great traders look to the right."
  • #10
  • Feb 27, 2009 1:19pm Feb 27, 2009 1:19pm
Quoting Ronald Raygun
This is definitely something which will require dynamic arrays of sorts to hold the tick, direction, and time data. I was thinking 3 parallel arrays of sorts:

TickValue[]
TickDirection[]
TickTime[]

Each tick is assigned a value, direction, and time. The problem I'm having is working with arrays in MT4. How do I adjust the size of the array without losing the internal data?
Ronald,

you can use ArrayResize( object&array[], int new_size)

as long as the new size is larger than the current size, you won't lose any data in the original array.

e.g.

int testArray[4] = {1,2,3,4};
// testArray holds 1,2,3,4

ArrayResize(testArray,8);
// testArray now holds 1,2,3,4,0,0,0,0

hope that helps.

regards,
Zen
Programming for a better future.
  • #11
  • Feb 27, 2009 3:11pm Feb 27, 2009 3:11pm
I've cooked up something based on the requirements by threadstarter. All the calculations and arrays manipulations are done. As trading is closed, I can only eyeball the code logic and it looks ok. I don't have time to put in alerts and/or trading functions as I need to get back to more pressing projects. Perhaps you guys wanna build up using this or take it as reference. PM me if you need me to explain my codes. I'll pop by when I'm taking a break again.

regards,
Zen
Attached File(s)
File Type: mq4 The_Possible_EA_v1.mq4   4 KB | 915 downloads
Programming for a better future.
  • #12
  • Feb 27, 2009 4:37pm Feb 27, 2009 4:37pm
Thank you very much! Two days ago it was just an idea I had, and today it became something real.
" Average traders look to the left, great traders look to the right."
  • #13
  • Edited 6:25pm Feb 27, 2009 6:14pm | Edited 6:25pm
I've spent the past hour testing this EA. Very, very nice! Another really big thanks to Zen Leow!

Testing is difficult because it's not programmed to leave behind and kind of marker (arrows, stars, etc.) when specified levels are reached. This means you have to test it slow enough for your eyes to catch the little numbers changing and still have time to pause the tester and leave notations on the chart itself, like I've been doing. My eyes are really tired about now.

My first observation is that the ticks reach the 65/35% level barely 10% of the time. Would any one else have guessed that? Not me. I would have guessed more like 25-30% of the time.

The best results (overall trading activity) seemed to occur when the ratios averaged UNDER 60% (53-56% seems best), and when OVER 70%. The 10% between 65-75 seems to be ranging territory.

Here comes the most interesting part: The lower ratios signified reversals, while the higher ratio (above 75%) signified same direction trends.

Pic1 shows the lower percentage reversals, and pic2 shows the higher percentage trending. The single red arrows in pic1 means it just reached the 65% level, but still less than 70%. The multiple arrows in pic2 mean 75% or higher. The one candle with 3 arrows was 85-87%, the highest of the test.

So far the results seem really promising. There is A LOT of usable data hidden inside these numbers that we can take advantage of. For more reliable testing automatic markers need to be added. Would anyone be able to help out with this?
Attached Image(s) (click to enlarge)
Click to Enlarge Name: test1.gif Size: 25 KB
Click to Enlarge Name: test2.gif Size: 24 KB
" Average traders look to the left, great traders look to the right."
  • #14
  • Feb 27, 2009 9:58pm Feb 27, 2009 9:58pm
Zen_Leow,

You are great. Your code is quite efficient and precise.
I respect you.

I found several misunderstandings about thread starter's intention,
so I deleted mine.
It's just confusing.

Regards.
  • #15
  • Feb 28, 2009 8:42am Feb 28, 2009 8:42am
This indicator shows amazing potential, and I'm not just saying that because it was my idea.

This has HUGE potential. This is not based on price action, this IS price action.

For better and more thorough testing it needs trade functions added on. It needs to be able to leave behind markers of some kind (arrows, stars, etc.) so that it can be tested a full speed.

Since this is based 100% on PA, combining it with lagging indicators (RSI, MACD, etc.) would render it less efficient. For best results I think it needs to be combined ONLY with other PA trading styles, such as pattern recogniton (IB, Dojis, PIN's, etc.)

Big props to Zen Leow for bringing us this far on such incredibly short notice, but is there anyone else willing to help take it to the next level?
" Average traders look to the left, great traders look to the right."
  • #16
  • Feb 28, 2009 11:41am Feb 28, 2009 11:41am
Quoting magicchance
This indicator shows amazing potential, and I'm not just saying that because it was my idea.

This has HUGE potential....

I can't code but I agree entirely with what you say ... this looks very promising - what are you looking for to improve it?

I don't trade with EA's but I think this would a great indicator for scalping using PA.

If it could mark off short term support and resistance levels it would be a great tool to use for reversals/breakouts.
  • #17
  • Feb 28, 2009 12:27pm Feb 28, 2009 12:27pm
Quoting lohikeitto
Zen_Leow,

You are great. Your code is quite efficient and precise.
I respect you.

I found several misunderstandings about thread starter's intention,
so I deleted mine.
It's just confusing.

Regards.
Thanks for the compliments. It was nothing, cos I did something a little similar sometime ago for someone else. If not for confidentiality agreement, I would've posted that one up for you guy's reference.

regards,
Zen
Programming for a better future.
  • #18
  • Feb 28, 2009 12:29pm Feb 28, 2009 12:29pm
Quoting magicchance
This indicator shows amazing potential, and I'm not just saying that because it was my idea.

This has HUGE potential....
How/where/when do you want the markers to be drawn? We should do this one step at a time. first to mark things out.
Only plant in the trading functions after the signals are correctly coded.
Programming for a better future.
  • #19
  • Edited 2:25pm Feb 28, 2009 12:35pm | Edited 2:25pm
WG - To answer your question let me ask one of my own: What exactly is volatility? Once that can be answered we can go further and ask, what is High and Low volatility?

The subject of volatility is brought up 1000 times per month here at FF, but I haven't seen anyone accurately define it. Describe it vaguely, sure. But not define and extrapolate it. The best attempts I've seen is the creative use of certain indicators, like BB for example, where a tight channel implies the opposite of a wider one.

We all know what PIN's, Doji's and IB's are. But do we know what makes them up? What exactly are the bits and pieces of PA that form the fundamental structure of these various patterns?

So these are the type of questions that brought me to this point.

My hunch is that programs such as this one can tell us what we need to know about the key features that make up volatility and patterns.

When it comes right down to it, what everyone here at FF is looking for, the magic bullet or Holy Grail, is nothing more than some type of filtering system. Some EA, some indicator, some oscillator, some type of pattern recognition, etc., etc., that helps to filter out more of what we don't want to see or be bothered with. Conversely, why not create something that shows us more of what we DO want to see as well?

To answer your question, going strictly by mechanics, I'd like to see Zen Leow's central program surrounded by a stand-alone EA with full trade capabilities and pattern recognition, at least for the 5 or 6 most common patterns. In another programming thread Hanover has offered such a program recently. Ronald Raygun has been updating his Generic EA with many good trading features. Why not combine all 3 and see what we get?
" Average traders look to the left, great traders look to the right."
  • #20
  • Feb 28, 2009 12:43pm Feb 28, 2009 12:43pm
ZL, sorry for that but I was writing while you were posting.

For testing purposes (prior to a working trade version) the EA should leave markers (arrows) at every percentage point reached, as set by the user.

Meaning that if I set up 65/35, then an arrow would be left at every tick where this % occurs.

For best results it would be absolutely fantastic to have four such windows to fill in so we could could keep track of 4 different percentage crosses at a time. So that for example if it's set up as 55/45, 58/42, 61/39, 77/23, it would leave a different colored arrow at each crossing point for easy visual referencing.
" Average traders look to the left, great traders look to the right."
0 traders viewing now