Optimized Trend Trading | Page 52 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Optimized Risk vs Reward Equation 335 replies

Loss Less Optimized Strategy 63 replies

Optimized Parameters in Metatrader Testing? 3 replies

"Optimized Parameter" feature in EA? 0 replies

Strategy with Optimized parameter 1 reply

Attachments: Optimized Trend Trading
Exit Attachments

Optimized Trend Trading

  • #1,021
  • Feb 11, 2011 7:06pm Feb 11, 2011 7:06pm
[IMG]file:///C:/Users/destiny/AppData/Local/Temp/moz-screenshot.png[/IMG]Yep

I don't know if I go correct release of the J_TPO indicator but it shows quite same thing than the last indicator LinearRegSlope

[IMG]file:///C:/Wutemp/I_dunno.gif[/IMG]
Attached Image (click to enlarge)
Click to Enlarge Name: I_dunno.gif Size: 21 KB
  • #1,022
  • Edited 9:05am Feb 16, 2011 9:03am | Edited 9:05am
Quoting John Last
http://en.wikipedia.org/wiki/Runge's_phenomenon

In my humble opinion there is an inherent problem using high order of polynomials. Take a look at this article.

I prefer by far to use the Mladen casual end - point SSA. Save your time use what is best.
Hi John

Is your affirmation true also for EmpiricalModeDecomposition ?

I just modified the first one, by putting inside the kernel mode calculation in empiricalModeDecomposition-Work4.

Rgds
john


BTW, do you know where to find jobs for trader in Switzerland ? I am just a Meta Trader administrator... oops
Attached File(s)
File Type: pdf Epperson329-341.pdf   735 KB | 1,382 downloads
File Type: mq4 EmpiricalModeDecomposition.mq4   5 KB | 1,279 downloads
File Type: mq4 EmpiricalModeDecomposition - Work4.mq4   8 KB | 1,307 downloads
  • #1,023
  • Edited 10:50am Feb 16, 2011 10:22am | Edited 10:50am
It irks me to no end when good people chase their tails as the result of someone else's work passing for something it is not. I recall when Ehler's article came out. It was of particular interest to me because I had been studying for some time Empirical Mode Decomposition. As I later found out, Ehler's Empirical Mode Decomposition bears no resemblance to the actual Empirical mode decomposition. If you follow the code it is merely a phase adjusted average based on the SMA. It doesn't decompose the data in any way, nor does it identify a trend any better than a phase adjusted SMA does. The real Empirical Mode decomposition does a good job of decomposing the data to isolate the various waves or trends, but is virtually worthless for trading due to the instability of the endpoints.

Quoting jaguar1637
Hi John

Is your affirmation true also for EmpiricalModeDecomposition ?

I just modified the first one, by putting inside the kernel mode calculation in empiricalModeDecomposition-Work4.

Rgds
john


BTW, do you know where to find jobs for trader in Switzerland ? I am just a Meta Trader administrator... oops
  • #1,024
  • Feb 16, 2011 11:08am Feb 16, 2011 11:08am
Quoting FXEZ
It irks me to no end when good people chase their tails as the result of someone else's work passing for something it is not. I recall when Ehler's article came out. It was of particular interest to me because I had been studying for some time Empirical Mode Decomposition. As I later found out, Ehler's Empirical Mode Decomposition bears no resemblance to the actual Empirical mode decomposition.
Hi

In fact, someone by private mail, asked me to put some polymonial calculation inside the code in the indicator I posted

You got the both codes for comparaison.
I have no understanding regarding this kind of algorythm. May be you can put in attachment, the Ehler's article and I will try to code it as Ehler would like to see it.

Rgds
  • #1,025
  • Edited Feb 17, 2011 3:39am Feb 16, 2011 12:31pm | Edited Feb 17, 2011 3:39am
Quoting jaguar1637
Hi

In fact, someone by private mail, asked me to put some polymonial calculation inside the code in the indicator I posted

You got the both codes for comparaison.
I have no understanding regarding this kind of algorythm. May be you can put in attachment, the Ehler's article and I will try to code it as Ehler would like to see it.

Rgds
What I'm saying is that the original work done by Ehlers is a fraud, a fake and doesn't do Empirical Mode Decomposition (EMD) - doesn't decompose the data and doesn't identify trend according to the scientific method developed by N. E. Huang, contrary to his claims. The indicator is a simple average with some DSP to make it appear fancy. I provide more info below but I wouldn't waste your time on Ehlers' junk science.

This link contains EasyLanguage code for the indicator (looks about the same as the original)

http://www.tradecision.com/support/t...omposition.htm

Here is an excerpt from the original article (can't find the original)
http://www.traders.com/Documentation...03/Ehlers.html

My theory is it just provided Ehlers a way to publish another article, so he can push his software / services. It's not even a very good forgery in that it makes no attempt to replicate the real Empirical Mode Decomposition.

As for the original work look for articles by N. E. Huang - there are many papers on the subject. This is a decent overview of EMD strengths and weaknesses and a comparison of other spectral methods.

http://keck.ucsf.edu/~schenk/Huang_etal98_review.pdf

Edit: Here is the original article in pdf form: (If you look closely, most of the code is for a bandpass filter, the actual attempt at trend decomposition is a miserable failure.)
http://www.mesasoftware.com/Papers/E...OMPOSITION.pdf
  • #1,026
  • Feb 17, 2011 1:40am Feb 17, 2011 1:40am
Can anyone guide me, how can I benefit from detrended fluctuation analysis in layman's guide?There are plenty of thesis but they are hard for me to understand for now, how does it related to hurst exponent?One of the paper I was reading says :

We find that all the scaling exponents are between 0.5 and 1 by DFA method, which exhibits the long-range power-law correlations , what does it mean?Thanks
  • #1,027
  • Edited 5:59am Feb 17, 2011 5:38am | Edited 5:59am
Jaguar,

I quite agree of what you think and your upgrade est vraiment exceptionnel and excellent but I disagree with your actions.
If you look inside you will would note that this indicator belongs only to the elite section in TSD.

It is an infringement. You should not do it.
For those reasons I can discuss this indicator in PM only.

It will benefit to our community if we share some kind of respect for each other and especially for the developers who have contributed most giving us for free excellent tools. For example most of the systems on TSD rely severely on the free ZerolagMA of Mladen.
Do not offense such kind of guys this does not benefit to nobody, and the people on FF and TSD we are mostly the same.

FXEZ I also wondered quite a lot why Ehlers called this indicator Empirical mode decomposition. Anyway he is my hero and I would not make any kind of critics .

I hope it is possible to delete the TSD indicator and somebody could make a different mql adaptation, because the formula is free and this is not a "secret" indicator just we are not allowed to use this code, for ethical reasons mentioned above. The code should be different.
  • #1,028
  • Feb 17, 2011 6:18am Feb 17, 2011 6:18am
Quoting John Last
Jaguar,

I quite agree of what you think and your upgrade est vraiment exceptionnel and excellent but I disagree with your actions.
If you look inside you will would note that this indicator belongs only to the elite section in TSD.

It is an infringement. You should not do it.
For those reasons I can discuss this indicator in PM only.
Ooooooops !
John, I did not realize this EmpiricalDecompositionMode.mq4 indicator was belonging to the Elite Forum, even it was indicated on top of this indicator. I got this indicator from a PM from another forum and the algorythm shown from the following URL :
http://www.traders.com/Documentation...adersTips.html

If you directly click on the
Tradestation: Emipirical mode decomposition, the algorythm is fully explained, w/o any reference to the Elite Section, just an Ehlers'article.

As FXEZ wrote after, this algorythm is just a simple average with some DSP.

I will write you in PM to fully understand this kind of situation.

Rgds
  • #1,029
  • Feb 18, 2011 1:09am Feb 18, 2011 1:09am
Anyone still running with this?
  • #1,030
  • Edited 3:28pm Feb 18, 2011 3:01pm | Edited 3:28pm
Quoting MisterH
Somebody please explain me the use of non-causal indicators. And no, smoothing the end-point with the JJMA doesn't work.
Now we have an end point version of SSA.

/All the SSA based indicators should be updated with the end - point casual SSA mod/

And you can see clearly that non causal indicators are of some use.

If SSA was not of some use I would not think it would be used in very complex climatology models casual or not casual.

Everything depends on how you use it, as a holy grail black box system,or with variable degree of understanding.


2. Smoothing the end-point with the JJMA does work. And it works pretty well but it is not a holy grail magic either. And I encourage to use it as much as you can.

And now to the general audience.

People keep looking for the holy grail systems that do not exist.
What we should strive is a sound principle based system with a positive expectation. This is not the same.

OK please how can a system keep working when the underlying fractal structure of the time price series varies. This is a fundamental feature of the markets (fractality, and the fractal bubbles).

OK let see, we have persistent market conditions and it works good as a directional algorithm.

After that you have anti-persistent market conditions (the probability that the price would reverse is greater that the continuation probability)

So how can you expect that the system would work equally well all the time. And that is the reason n 1 why back testing and forward testing do not work as good we want , that is why I prefer not a statistical optimization based on a historical period but a phenomenological optimization, we are at a great disadvantage we miss high frequency data to properly optimize anything.

And occasionally people do not even realize when their system does give a solution in a particularly difficult market conditions /expecting it to give tops and bottoms nothing less (yes it is possible but at very specific circumstances of low phase space singularity as today for example, many systems did). This is foolish. And there is a constant phenomenon of system migration.

It is ugly to say but many come with the following mentality. They come to the PC at a discretionary time. Turn on the MT4 and expect that the system will give them the right entry. After that they want some money, right now no matter what the market conditions are.

This is an extreme example no 2.

Others think that because they are cleverer and have coding skills and can set up a complex artificial intelligence model they can beat the market any time, anywhere.

No we cannot beat the market we just want to risk our money when the odds are in our favor.

And when the odds are in our favor the game is not to predict anything, the game is to fish in a barrel.
  • #1,031
  • Feb 19, 2011 10:22am Feb 19, 2011 10:22am
Quoting John Last
Here I add my alma mod of the BB Caterpillar Squeeze on Alma.

I particularly like this mod. This time the Histogram is based on Alma and the dots are the Caterpillar squeeze. I used the open price for the Caterpillar, as I do not want it to recalculate for every tick, that can hurt some weak PC.

I also add the normal Squeeze with Alma. In the normal mod you do not have any Caterpillar (SSA). You have the normal squeeze with green bars. Look at the video to see how it has to be used.

How you should use it. Well the green dots of the normal...

Hello John,

I have downloaded your indies, but they do not show on my chart. Any help is appreciated. The bbsqueeze alma shows on the chart, but it does not look thesame as yours.

Thanks
  • #1,032
  • Feb 19, 2011 12:24pm Feb 19, 2011 12:24pm
Quoting John Last
Now we have an end point version of SSA.

/All the SSA based indicators should be updated with the end - point casual SSA mod/
Well, john

I would like to talk about an old indicvator, well known the Price Volume Trend, based on this following calculation
_________________________________________
We get PVT by multiplying the current volume by the relative change of the share price and adding the result to the current cumulate value of the indicator.
PVT (i) = ((CLOSE (i) - CLOSE (i - 1)) / CLOSE (i - 1)) * VOLUME (i) + PVT (i - 1)
Where:
CLOSE (i) — the closing price of the current bar;
CLOSE (i - n) — the closing price n bars ago;
VOLUME (i) — the volume of the current bar;
PVT (i) — the current value of PVT indicator;
PVT (i - 1) — the value of PVT indicator on the previous bar.
__________________________________________

Do you think the SSA calculation could be the next improvment inside this indicator ??

Rgds
Attached File(s)
File Type: mq4 PVT.mq4   3 KB | 1,093 downloads
  • #1,033
  • Edited 5:22am Feb 20, 2011 5:04am | Edited 5:22am
Unfortunately I limit the number of projects I participate. What I want is not a thread with hundreds of posts ( almost impossible to read). I want a thread with a manageable number of posts with a complete but open to evolution system.

There are too much indicators. When you apply different kind of mods their number increases exponentially. What really count after all is the system that is build around them.

The best I can offer to our community is the Digital ASCtrend (signals) with a combination of Brain trend with SSA end point (stops and signals).

http://www.forex-tsd.com/manual-trad...-acstrend.html

For slow filters there are some very interesting things.
Those are the three filters in my opinion by now, that I can recommend.

-MESA adaptive filters: FATL and SATL
-SSA end point filters and SSA end point with adaptive normalization period
-And recently a jurik filtered entropy indicator

(those indicators are not collinear with the standard indicators we do not talk about incremental improvement, they are simply different)

/SSA squeeze and the BBMACD are very good together with Trend magic mods however it is a question of a personal choice and experience/

I add to the collection and the entropy indicators allowing to gauge the market conditions together with the fractal dimension indicators. For example we look for drops in the entropy. Very often they are an early warning for a technical break out.

/The market condition template would be an independent project. This template would combine non directional indicators. For example the volatility indicators are widely used measuring the volatility: JCFBaux (my favorite), Damiani Volatility indicator, Squeeze of moving averages, Bollinger bands, ATR etc.

However we can add:
Fractal structure indicator: FGDI, FDI, iVAR: fractal dimension and break outs

And lately Entropy indicators: entropy break outs and entropy drops

We miss only as a free indicator the Ehler's noise to signal ratio, there is only one implementation for MT4 but it is not free and I cannot share you in this forum./

Here the indicators are just prototypes inviting you to build your own system or insert them into the existing systems in FF allowing for some variation (with the reserve of proper back testing of course). The choice is yours.

Please be more specific which indicator you cannot run.
And most of the indicators in this thread using SSA are outdated by now, you need to replace the ssa (caterpillar) with the Mladen's SSA end point.

So somebody, anybody, can do it, I really do not have the time.

Jaguar please beware the volume you have in your MT4 has nothing to do with the real volume. This is not VSA (volume spread analysis) this is TSA (tic spread analysis) (it appears to me that the tic volume makes patterns on its own that should not be underestimated).
  • #1,034
  • Feb 20, 2011 11:41am Feb 20, 2011 11:41am
Quoting jaguar1637
Well, john

I would like to talk about an old indicvator, well known the Price Volume Trend, based on this following calculation
_________________________________________
We get PVT by multiplying the current volume by the relative change of the share price and adding the result to the current cumulate value of the indicator.
PVT (i) = ((CLOSE (i) - CLOSE (i - 1)) / CLOSE (i - 1)) * VOLUME (i) + PVT (i - 1)
Where:
CLOSE (i) — the closing price of the current bar;
CLOSE (i - n) — the closing price n bars ago;
VOLUME (i) — the volume of the current...
Hi Jaguar,

Then we're moving to a whole new area of exploration. Start here .
I believe version 6 is the easiest to work with.

Personally I would like to see if anyone can fit SSA into this, and the end-result, if any.

Good luck!
  • #1,035
  • Feb 21, 2011 3:19am Feb 21, 2011 3:19am
Quoting John Last

Jaguar please beware the volume you have in your MT4 has nothing to do with the real volume. This is not VSA (volume spread analysis) this is TSA (tic spread analysis) (it appears to me that the tic volume makes patterns on its own that should not be underestimated).
For what it's worth, I've found that the metatrader volumes are a useful way of deducing the overall activity in the market. Price movements are most extreme when the markets are open (Duh!), and the metatrader volumes *roughly* reflect that. Hence It's useful to add to a calculation formula. Just be very away that it's as rough-as-rats! It would probably be better to use a clock-time filtering method.

I've been following this thread for a while, and hope to be able to contribute some programming time towards helping out. Any suggestions (John?) where It would be most useful to concentrate first? I was thinking I might start with a free Ehler type S/N indicator.

It's been a while since I programmed MT4, and I hope that most of the bugs have been fixed. It's a painful platform to programme.

Flea
  • #1,036
  • Feb 21, 2011 5:17am Feb 21, 2011 5:17am
Quoting ElephantFlea
For what it's worth, I've found that the metatrader volumes are a useful way of deducing the overall activity in the market. Price movements are most extreme when the markets are open (Duh!), and the metatrader volumes *roughly* reflect that. Hence It's useful to add to a calculation formula. Just be very away that it's as rough-as-rats! It would probably be better to use a clock-time filtering method.
Flea
Hi Flea

Which algorythm would you like me to insert and adapt to the indicator.

rgds
  • #1,037
  • Feb 21, 2011 10:25am Feb 21, 2011 10:25am
Some experiments.
Attached Image (click to enlarge)
Click to Enlarge Name: ssa window.gif Size: 27 KB
  • #1,038
  • Feb 22, 2011 4:48am Feb 22, 2011 4:48am
Quoting John Last
Some experiments.
Hi John

I can see two indicators SSA normalized giving different drawings. Is it a different settings ?
Can you tell me more?

Regards
  • #1,039
  • Edited 7:34am Feb 22, 2011 5:58am | Edited 7:34am
Quoting ElephantFlea
For what it's worth, I've found that the metatrader volumes are a useful way of deducing the overall activity in the market. Price movements are most extreme when the markets are open (Duh!), and the metatrader volumes *roughly* reflect that. Hence It's useful to add to a calculation formula.
Flea
1). If metatrader volume behaves like real volume e.g: Volume precedes price moves.

2). It should also (I think) be reasonable to expect that (high - low) = k*volume^h. (Roughly)

Estimate k and h (k and h are constants, h < 1.0) using historical data and we may have a predictive "trend" / "no trend" (big / small bar) filter. That is if (high - low) less then predicted, a trend should be pending in the near future.

/LV - Just thinking.
  • #1,040
  • Feb 22, 2011 8:04am Feb 22, 2011 8:04am
The second shot is Volume adjusted SSA. When we have low tic volume (market activity) it looks like it sleeps.

Quote
double ma = iMA(NULL,0,SSAPeriodNormalization,0,MODE_SMA,SSAPrice,i);
double dev = iStdDev(NULL,0,SSAPeriodNormalization,0,MODE_SMA,SSAPrice,i)*iVolume("NULL",0,i);

double price = iMA(NULL,0,1,0,MODE_SMA,SSAPrice,i);

no[i] = (price-ma)/(MathMax(dev,0.000001));


Here I have added the volume to the calculation.
Normally it was just the standard deviation, but I added the volume as suggested by Camaron ,just to see what would happen. Anyway I am not sure if that is right.

Mr. LastViking from where you have this knowledge? From where this formula comes from? Why you believe that those should be constants and not variables? So maybe you would use the genetic optimizer to derive them!

Sorry for the questions.
0 traders viewing now