Trading Strategies Based On Digital Filters - page 87

New comment
 
Pava:
small joe...that was uncalled for.....

Damn, sorry Pava, thought that was you that claimed for the repainting fact

 

:)

Big Joe:
Damn, sorry Pava, thought that was you that claimed for the repainting fact

HeHe...I could but this time I stayed back and was watching:)

 
leledc:
Man is better...

Noted

 

...

Some interesting reading Does it "repaint" : I will let the "repainting" experts decide on that.
PS: a site with a wealth of books. No warez. No copyright violations. Imagine : Cornell University Library (here : arXiv.org e-Print archive ) The document attached comes from there. Maybe helps someone clear the "repainting" dilemmas. As far as I am concerned, when talking about Fast Fourier transform and similar "things", talking about "repainting" has as much meaning as it would to be ask "is 1+1=2?"
 

...

Simba,

I don't need a proof. At least i don't

But the usage of "repainting" when talking about mathematical models and methods is nothing but a waste of time. I just wanted to remind : asking over and over the same question without knowing what the basis of a particular system, or a model, or an indicator is meaningless. It does not show knowledge but a lack of imagination

And I am going to repeat what I already told in the elite section (long time ago): it is not repainting what we are talking when we are talking about these things. These are recalculations. REPAINTING IS A CODING ERROR. Calling, implicitly, FFT, SSA, whatever similar ... a coding error, is meaningless

regards

mladen

 

Proof

mladen:
Some interesting reading Does it "repaint" : I will let the "repainting" experts decide on that.
PS: a site with a wealth of books. No warez. No copyright violations. Imagine : Cornell University Library (here : arXiv.org e-Print archive ) The document attached comes from there. Maybe helps someone clear the "repainting" dilemmas. As far as I am concerned, when talking about Fast Fourier transform and similar "things", talking about "repainting" has as much meaning as it would to be ask "is 1+1=2?"

Mladen,

I consider you a friend and I appreciate alll the help you have given me during the past years,not to say anything of the help you have given everybody and his uncle here at this forum,So...with all my respect ....

Do you need a mathematical proof of trends?do you need a mathematical proof of anti persistence?...I don`t think so,since you posted it ...You know as well as me that price series can be in one of these 3 states...persistent(trending,positive feedback loop,higher prices lead to higher probability of higher prices),antipersistent(ideal for cycles,negative feedback loop,higher prices lead to higher probability of lower prices,ideal for cycles) or random...It is just a matter of avoiding the third one and using the right tools fo the other 2...Hurst exponent works marvels for daily or weekly data,IMO,for h4 and below timeframes,Hurst exponent doesn`t wiork,probably due to Signal to noise ratio.

BTW,I posted a mathematical proof of trends too,in another thread,I referenced it here,at this thread, a few months ago,when I was exchanging ideas with MadCow.

All other members,

Fourier:I will try to explain as simple as I can....

1-Does it repaints/redraws?...YES,BIG TIME.

2-Is it useful?..experiment with it and decide for yourselves....IMO,it is VERY useful.

3-According to theory ,and,for once ,to practice..price series are non stationary...so,applying Fourier to them wouldn`t make for anything useful...YES....OK,you probably didn`t realize it,so,I will repeat...IMO,applying FFT to price series has no edge ...please keep reading before committing seppukku.

4-You can take the first derivative of price,or the second until you find stationarity(you will find it)...OR...You can apply Fourier to something which you know is stationary...Is RSI stationary? ...Does it returns to the mean?...You can bet on it....is any normalized indicator stationary?..Of course,so,use Fourier on stationary data that models the "limits" of price....tips:RSI,Momentum,volatility...this,plus a robust strategy,like the 2 ones I explained,and sound MM will give you an edge...that`s all....This is a probability game,not a Holy Grail search.

ALL DATA IS STATIONARY IF PROCESSED WITH THE RIGHT TOOLS....So,process the data and use FFT on the processed data,which will be stationary by definition.

Pava:

Thanks for your indicator suggestion of a few days ago...today its FFT repainted big time,even so,the results have not been too bad,and the cumulative results of the past 3 days have been positive...My suggestion...the FFT indicator posted by Beppi a few posts ago,plus your indicator,make for a very good combo,I have been testing them at m1 and m5,and I am impressed by their capability to keep you in the right direction and scalp most of the moves...still more days to test,but it looks very promising.....

Regards

S

 

...

as far as I am concern...cycle indicators of the late threads show what market suppose to do in a perfect world...so I wait for confirmation...will post indicator after will get home....

 

Simba ,

I like the way you go on about ''everybody and his uncle''

 

hello

Pava:
as far as I am concern...cycle indicators of the late threads show what market suppose to do in a perfect world...so I wait for confirmation...will post indicator after will get home....

and are they useful?

 
mladen:
Some interesting reading Does it "repaint" : I will let the "repainting" experts decide on that.
PS: a site with a wealth of books. No warez. No copyright violations. Imagine : Cornell University Library (here : arXiv.org e-Print archive ) The document attached comes from there. Maybe helps someone clear the "repainting" dilemmas. As far as I am concerned, when talking about Fast Fourier transform and similar "things", talking about "repainting" has as much meaning as it would to be ask "is 1+1=2?"

Thanks for site, bookmarked, definitely a wealth of information!!

1...808182838485868788899091929394...138
New comment
Reason: