the impossible EA?

Here's v2 which comes with markers.

you don't need to define all 8 numbers for the ratios. Since it's percentage, you'll just need to state one and it'll imply the opposite figure. You do however need to define 8 different colors for the arrows to be drawn.

regards,
Zen

Do you mean 8 arrows (colors), or just 4? We only need one arrow per % pairing. Fours % inputs would need only 4 arrows.

Nevermind that last question. I get it now. You've programmed it to show the same percentage in both Bull & Bear directions. Very cool! Am testing it right now.

Zen Leow,

The tester works fantastic! It loads up and works just fine. Thank you very much.

But I have a question about the seconds filter. The idea is that this filter limits the amount of data being considered, so that the resulting arrow indicator only alerts when all three filters have been activated:

Filter 1: 'X' Ticks
Filter 2: 'X' Elapsed Seconds
Filter 3: 'X' Percentage (of Bull vs. Bear)

Filters 1 & 3 seem to be working just right.

But during testing I have been severly limiting the data to see where the extreme parameters are. In other words, I have been putting the '%' settings at 93/7, with the 'seconds' set at 100. In doing so I am trying to find the outer edges of the effectiveness of the EA. By using such extreme settings, the amount of '% level hits' should be almost nothing. I should be getting just a few hits per test run. But the thing is, I am getting just as many hits as I do when set for 250 seconds.

Filters 1 and 3 seem to be OK. That leaves Filter 2. It does not seem as though Filter 2 is limiting data accumulation. It seems to be acting only as a timer, but this is not what we need it to do. The purpose of this filter is to severly limit the amout of data the EA is to use, by "forgetting" all data older than what the timer is set for.

I apologize if I am going too long about this. But since we can only type out our thoughts, and not speak them, I just want to make sure we understand each other.

testing this ea on hystorical data would be worthless. the ea is based on ticks and when testing in mt4 the ticks are not real ticks but are being modeled. since for this kind of ea having precise and accurarte tick data is crucial testing (in mt4) can only be done live in order for the results to have any significance.
not critisizing the idea (i think its a great one) by any means just trying to make sure everyone understands how mt4 works and how that would affect testing this kind of ea

OK, so does this mean that even if the codes are exactly what I asked for, the tester would easily confuse me into thinking it the other way around? Because that's where I am with this. And Zen Leow has been so great that I want to make sure he undertsands I am not in any way criticising his work. But is this what I'm seeing you think?

Let me try to explain. The lowest "real" data increment in mt4 is 1 min. when testing, mt4 takes those 1 min chunks of data and simulates ticks up/down within that one minute bar. those simulated ticks are nothing more than guesses. so for example if you have a 1 min bar where the price has moved 10 ticks and all of those ticks have been up the tester could simulate that as 6 ticks up then 4 ticks down then 8 ticks up. that would mess up the tick count for the ea. for that candle you should have had 100% up but in testing you would have 18 ticks total 14 up and 4 down which is up 77% and down 33%. so when dealing with ticks this would have a huge impact on accuracy. IMO the only way to accurately test this is live. Once again this has nothing to do with the code or idea but with the limitations that mt4 has when it comes to ticks.
HTH

Got it!

Filter 2 is working fine. even during the highly inaccurate MT4 backtesting environment, it's showing those imaginary "ticks" don't get past the designated number of seconds before being purged.

another proof that Filter 2 is working would be to look at the "Current Index" value, that value tells you how many ticks are currently stored. you'll find that sometimes it drops a significant number. that's the expired ticks purging at work.

you can try setting the time to 50 seconds or even 5 seconds. you'll get a lot less arrows because that's too short a time to accumulate the bull/bear percentages. Also you'll be able to see the purging more distinctly.

Do I need to explain exactly how this purging mechanism works?

one more thing to point out. whenever a prescribed level is being reached, it's not uncommon for a lot of arrows to be drawn around that area, mainly because of price fluctuation and the percentage can go out and back to the same level repeatedly in a short time spent.

but still, just as sergiu has pointed out, only live forward testing can be truly accurate.

regards,
Zen

Got it, and got it again! LOL

That's the best thing you could have told me, and I am looking forward to the coming week so I can test it live! I plan to give this whole next week to running this EA on several pairs and try to collect some data!

Please check your PM box for a note from me.

Chance

nice work! The only area that is questionable is the significance of knowing the results of 4 exact percentages when 96 percentages are not considered.

ie: if (BullPercent == DivergencePercent_1)

KD, Not sure what you're trying to say. Mind trying again with deeper expanation?

Zen Leow,

As you can see by the pic, your program is working great. By now I have noticed a small bug I hope you can re-work, and then I have a programming request after that.....

The bug: Once a chart is active and there are arrows everywhere collecting data, should you ever re-set ANY part of the settings (tick count, seconds, etc.) all data is deleted. From that point all collected data is lost and it starts again from zero.

The request: Can you make it so that by toggling ALT/F1 on/off the first arrow set disappears & reappears? ALT/F2 for the second set, and so on to F8. Is this possible? Odd numbers are Bulls and evens are Bears.

(Less urgent improvements would be slightly larger arrows and larger print on the left.)

That's not a bug unfortunately.

Remember, tick data isn't stored by the terminal.

No, I'm not a newbie, honest!

well... then you shouldn't be using an EA. It should be an indicator.

EA will forcefully remove all the objects that it drew if itself gets removed or re-initialized.

Indicator doesn't remove objects when itself is removed or re-initialized (unless of cos if I code it to remove objects drawn, then it will).

Try this indicator instead. Its an almost exact replica with a few added bonuses (you'll literally SEE what I mean later). Remember, this one goes into expert/indicators folder.

regards,
Zen

Well, you definitely fixed the size problem! LOL

It's ticking away like crazy. I like it! My fingers are getting sore from typing "Thank You" so much, but I don't mind at all!!!

What are the last 4 settings for?

And is the 'F' key toggling idea possible, or did I finally find your Kryptonite?

Using hotkeys in metatrader can trigger a script which changes a global variable which enables/disables a function in the indicator....

Just my thoughts

I like your thoughts just fine, RR. Step in here when ever you can!

The thing I'm getting at with the F key idea, is that with 8 different colored arrows all over the place, the chart can get really crowded during periods of high volume (which is a good thing.) But rather than disabling the whole indicator (all 8 colors at once), how about something that will temporarily disable just one color (set of arrows) at a time? Being able to selectively remove (and replace) specific data at the touch of a key would be a big improvement.

Firstly This coding is super duper. You're lucky to have such nice work to start from! Of a thousand indicators, there are very few that use oop/objects, let alone may implement globals as RR has mentioned.

It's the logic that is flawed.

The line: if (BullPercent == DivergencePercent_1)

means that if the bulls percent is not exactly equal to this number then toss out the results. Your ea is looking for exactly 4 numbers out of 100. Why? maybe there is a reason, if not then its possible that your looking for needles in a haystack atm...

I would reccommend changing the percentages to filter for a range of results. something like:

if (BullPercent >= DivergencePercent_1)&&(BullPercent <= DivergencePercent_2)



Nice one. Where can i find OnObjectClick[] documentation


----------------------------------------------------------
All I know for sure is if I short EURAUD at 2.00 it will go up