Symphonie EA

[garyfritz]

In which way could your post be described as, "More pithy"?
Ok, so it may be more pithy, so my apologies. How about more pithy and something I can actually use in the EA?

Hey, I thought

BaseMag = MathPow(10, MathFloor(MathLog(Acct) / MathLog(10))); // 4376.59 -> 1000
RoundDown = MathFloor(Acct / BaseMag) * BaseMag;

...was fairly pithy and reasonably understandable, given my explanation. And you CAN use it in the EA. I tested it in an EA and it works. You asked for mathematicians to look at it, so you got a mathematical answer. I'd say it's safer than your string-manipulation hackery...

Mahmut's solution only works for account balances in the range 1000-10,000. For balances < $1000 it returns $0, and for balances > $10k it returns e.g. $44321 -> $44k instead of $40k. (And if that's what you want, then great.) Mine works for any account balance: $42 -> $40, $532 -> $500, $2345 -> $2000, $12345 -> $10k, $43210 -> $40,000, etc.

[slowkey]

For lack of a market to forward test I spent some time looking at this EA in the Crap-t4 tester. Looking at specifically EURUSD. Looked at it on several time periods with MTF on. Looked at it a lot in visual mode. Found that the 60 minute time frame with the RSI 1440 / 20 filter on looked rather good. That was with MTF turned off. Optimized some setting just to zero in on something that might work on forward testing. My gut feeling is that Evaluator has spent a lot of time fine tuning his system on EURUSD. So I ran the Crap-t4 torture test on it trading 1 lot positions with the most aggressive setting and a 50,000 bank over 11 years. It survived with 30,000 left. I think what we got here is a EA that has been highly optimized to trade EURUSD. Certainly worth forward testing now on this time period. I will continue to test Evaluators MTF 60-5 on the laptop but am very encouraged to look at this as a EURUSD trader on the 60 minute time period with the filter. I will try several different setting. Strategy one with a 100 stop or exit by extreme stop only. Than a conservative 60 stop 90 target. Than two more aggressive setting which I am still looking at. Of course with a target we are either getting a extreme stop before the target to exit the trade or a target exit. Nothing means anything unless it forward test well too. I do think that an EA is more stable on the higher time frames. I still think that the symphnonie 60-5 may need a filter on the LTF something like an adx. I think though that Evaluator seems to have an excellent understanding of indicators and maybe able to help us with some ideas that would fit best in his system.

[Mediator]

P.S.: if you like it then please give me vouches at FF. The more vouches the more respect me at FF. Oh man, what a kindergarden.

Kindergarten :

So vouches can also be obtained for manufacturing dreams and not necessarily for creating value : :!:

Khalid

I did not really understand your answer becauseof my lagging english.
I forgot to switch on my irony-mode with my post about vouches. And i was angry that i was treated like an unknown child and did not getting respect that i shared my experiences.

[SteveHopwood]

In which way could your post be described as, "More pithy"?
Ok, so it may be more pithy, so my apologies. How about more pithy and something I can actually use in the EA?

Hey, I thought

BaseMag = MathPow(10, MathFloor(MathLog(Acct) / MathLog(10))); // 4376.59 -> 1000
RoundDown = MathFloor(Acct / BaseMag) * BaseMag;

...was fairly pithy and reasonably understandable, given my explanation.

Hehe. That is a mathematician talking. The rest of us had not a clue what you were talking about (ignores cries of "Speak for yourself, dimbo"). By the time I got to the above, I had lost the will to live and no longer understood even single syllable words. We all think that others should be able to master our skills; I can never understand why others find playing piano concertos so difficult.

This is fantastic, Gary. Many thanks. I did look at MathLog whilst trying to work this out for myself but did not understand. I sort-of- do now, having read your code. I understand MathFloor as well, something I did not yesterday. This is how I learn, by studying working examples and seeing how they tick. I recently learned about the invaluable MathMax/Min from TIG.

So, your code is added to S, with due appreciation in the comment.

Latest update in post 1 uses Gary's contribution. I too have tested it and it unbreakable, so I suggest that those of you wanting to use ARS do a redownload.

[garyfritz]

Maybe I assumed more understanding of the MathXXX functions than was warranted. Given the "pithy" nature of Empty4 documentation, it would be understandable if you didn't have a clear understanding of what those functions do. If it would be helpful, I would be happy to provide a quick summary of what they do, with examples, which might increase your comfort level with those functions.

[SteveHopwood]

Maybe I assumed more understanding of the MathXXX functions than was warranted. Given the "pithy" nature of Empty4 documentation, it would be understandable if you didn't have a clear understanding of what those functions do. If it would be helpful, I would be happy to provide a quick summary of what they do, with examples, which might increase your comfort level with those functions.

That would be fabulous, thanks.

I now understand several of the functions thanks to you and TIG. Others may lack even this rudimentary knowledge, so if you have the time to document the lot that would be brilliant. I can add it to my shell code thread for the benefit of all who pass by.

Cheers

[slowkey]

I did get a MTF time frame of 15-60 working on eurusd. It did not look as good as the 60 minute with just the rsi. The reason being that it looked to have at minimum double the drawdown. I will be testing this approach too. 4-1 time period relationships seem to work well in trading mtf setups.

[cuzgeorge]

Code: Select all

This is fantastic, Gary. Many thanks. I did look at MathLog whilst trying to work this out for myself but did not understand. I sort-of- do now, having read your code. I understand MathFloor as well, something I did not yesterday. This is how I learn, by studying working examples and seeing how they tick. I recently learned about the invaluable MathMax/Min from TIG.

So, your code is added to S, with due appreciation in the comment.

Latest update in post 1 uses Gary's contribution. I too have tested it and it unbreakable, so I suggest that those of you wanting to use ARS do a redownload.

Hi all, hope all had agreat week end and are well rested.
Thank you Gary for the input, you guys are geniuses,


Steve, i have a question for you, similar to the ARS issue; (coding i mean)

Your PM also explains why i keep getting this error in the 'experts' tab. I get it in all instances of the Meta traders i run. Every chart has this problem and when an 'alert' is trying to pop up, it can freeze that instance of Empty4 for a good couple of seconds and even up to a minute or two. i am now also testing MT5 and hoping to get better results, even though i still have to find a way to get my custom 'stuff' and symphonie onto it, and actually see it working in realtime.

2012.03.25 21:36:02 Cannot open file 'C:\Program Files (x86)\Ava Empty4\experts\indicators\ZigZag SG.ex4' on the EURUSD,M5




Similar to GaryFritz or Raygun, i cant remember , i use smaller TF's to get in and then revert to the larger TF's like 30M ish. i find larger than this is tooooo slow. i need action dammit. hehe.

This leads me back to symphonie and RSI. Symphonie has the potential to follow up and down, i like that, but it's the first time i feel like this can work. so far, i've done a lot of testing but have'nt looked at the code much. i was wondering about the section where you have a note that the RSI uses the 55, 45 settings. can this be changed? say to 67, 25? I'm also using 6=weighted setting at the moment.

Your comment, opinion please?


"//RSI. >55 is trending up. < 45 is trending down
string trsi="Rsi trend detection";
bool UseRsiTrendDetection=false;
int RsiTdTf=1440;
int RsiTdPeriod=20;
string trs="Applied price: 0=Close; 1=Open; 2=High";
string trs1="3=Low; 4=Median; 5=Typical; 6=Weighted";
int RsiTdAppliedPrice=0;"

I am still getting used to the settings, there are so many and rightfully so, i need to understand better to find the correct settings. You have done your part, now we'll do the running and report back on the forum.

Slowkey,
i have started to make use of FXbook and hope all will be able to see results without those silly pics i've been trying to get my results through to you all. I'm not done yet, so please be patient until all are up and running in the course of the week.
Please look at the 'comments' tab to get a brief idea of what settings i used. I f you would like the set file, please quote me which pair and from what result so i can find the correct set file for you. There are many settings and i'm following all up. i'm trying to not take up so much space on the forums, by doing this, it's kinof embarassing. hehe.

regards
cuzgeorge

[SteveHopwood]

George, the error means you do not have the required indicator in your indi folder. An EA cannot open a file that does not exist.

Re the RSI thingy, making changes to the trend-detection range is easy. Go to void TrendDetectionModule(). The Rsi trend stuff is the first code block. Change the values is this

Code: Select all

if (TrendRsiVal > 55) RsiTrend = up;
if (TrendRsiVal < 45) RsiTrend = down;

to whatever you want and recompile. Using the example in your post:

Code: Select all

if (TrendRsiVal > 67) RsiTrend = up;
if (TrendRsiVal < 25) RsiTrend = down;

[garyfritz]

If it would be helpful, I would be happy to provide a quick summary of what the [MathXXX functions] do,

That would be fabulous, thanks.

OK, here are all the MathXXX functions. Some are probably obvious but I'll cover them all.

MathAbs(X) -- absolute value of X. MathAbs(3) = 3, MathAbs(-3) = 3.

MathMax(X,Y), MathMin(X,Y) -- the largest or smallest of two numbers.

MathMod(X,Y) -- the "modulo" function, which returns the remainder of a division. E.g. 7 / 3 = 2.333, or "2 with 1 remainder." MathMod(7,3) returns 1 -- the remainder of the 7/3 division. Non-integer values work -- e.g. MathMod(8.3,2.5) returns 0.5, because 3*2.5 = 7.5, and there is 0.8 "left over" between 7.5 and 8.3.

MathRound(X) -- the nearest integer value. MathRound(2.1) = 2.0, MathRound(-2.1) = -2.0.

MathFloor(X), MathCeil(X) -- The closest lower or higher integer value. MathFloor(2.2) = 2.0, MathCeil(2.2) = 3.0. MathFloor(-2.2) = -3.0, MathCeil(-2.2) = -2.0.

MathSqrt(X) -- square root of X.

MathPow(X,Y) -- X raised to the Y power. So MathPow(2,3) = 8, because 2^3 = 8. Non-integral values work. MathSqrt(X) == MathPow(X,1/2).

MathExp(X), MathLog(Y) -- the constant e (~= 2.71828) raised to the X power, or the natural log (base e) of Y. (Don't ask why they use base e. It's a deep involved mathematical explanation. Imagine trying to explain Myxolidian mode to someone who's never heard music. ) These are inverses of each other, so if MathLog(A) = B, MathExp(B) = A. Since MathExp(X) is e raised to the X power, MathExp(X) = MathPow(e,X). Logarithms are easiest to understand if you look at base-10 logs. The Log10 of X is the value that satisfies 10^Y = X. So e.g. log10(100) = 2.0, because 10^2.0 = 100. Non-integral values also work, e.g. log10(4321) = 3.6356, because 10^3.6356 = 4321. MQL only supports natural (base e) log, but you can get log10 from MathLog: log10(X) = MathLog(X) / MathLog(10).

MathSrand(X), MathRand() -- Pseudo-random number generator. (Because software cannot generate truly random numbers.) MathSrand() starts or "seeds" the random series. You might feed it TimeLocal() to give it a nice semi-random start. MathRand() pulls the next random number in the sequence.

MathCos(X), MathSin(X), MathTan(X) -- sine, cosine, tangent of X. Probably not used all that much in EAs.
MathArccos(X), MathArcsin(X), MathArctan(X) -- the *inverse* of the above. So if Mathcos(X) = Y, then MathArccos(Y) = X.