The Sultonov system indicator - page 45

 
Unicornis:

Interesting, please give successive examples: which pair(s) and date-time of coefficients leaving the channels.

eurodollar pair, further, see page 1 of the thread

Desperate attempts by odds to exit the channel for a trend change, but, they fail to break the downtrend, although, there was a real threat:


 

I haven't read the whole thread. It may have already been mentioned.

Your "method" is a simple regression using the value of the next bar that has not yet arrived. Hence these "miraculous" results and "predictions". If you are going to sell such an indicator, it is an outright fraud.

 
Murad Ismayilov:

I haven't read the whole thread. It may have already been mentioned.

Your "method" is a simple regression using the value of the next bar that has not yet occurred. Hence these "miraculous" results and "predictions". If you are going to sell such an indicator, it is an outright fraud.

It's kind of like a collective grail search.

 
In fact, it's time for a real prediction. A few of these and the programmers will come running in, you'll get tired of waving them off, aftar!
 
Murad Ismayilov:

I haven't read the whole thread. It may have already been mentioned.

Your "method" is a simple regression using the value of the next bar that has not yet occurred. Hence these "miraculous" results and "predictions". If you are going to sell such an indicator, it is an outright fraud.

If the bar has not yet occurred, then how and where do I get the price of the bar that has not yet occurred for fraudulent use? You have used such a fraudulent phrase yourself that even Ostap Bender would be amazed! Just for your information, do I use the opening price of an OPEN bar, or is it forbidden to use it? Explain to us, the ignorant, from the height of your knowledge in the field of tautology, about the brightest of luminaries and the wisest of wise men!

 

the method is tested quite simply, you can even brute force it.

You take a harmonic function and "predict" its next value. You get that at some frequencies/amplitudes the "predictions" are better, at some worse. And if you add minimal noise, it is almost always a finger in the sky.

As a result, when there is a "good" component in the market, the function works, otherwise it doesn't. And the fact that the market carefully avoids repetitions and "squeezes" the resulting fluctuations is its nature, and it has been verified by many years of work by hundreds of analysts.



 
Maxim Kuznetsov:

the method is tested quite simply, you can even brute force it.

Take a harmonic function and "predict" its next value. You get that at some frequencies/amplitudes the "predictions" are better, at some worse. And if you add minimal noise, it is almost always a finger in the sky.

As a result, when there is a "good" component in the market, the function works, otherwise it doesn't. And the fact that the market carefully avoids repetition and "squeezes" the resulting fluctuations is its nature and it has been verified by many years of work by hundreds of analysts.



It's even easier - by coffee grounds!

 
Maxim Kuznetsov:

the method is tested quite simply, you can even brute force it.

You take a harmonic function and "predict" its next value. You get that at some frequencies/amplitudes the "predictions" are better, at some worse. And if you add minimal noise, it is almost always a finger in the sky.

As a result, when there is a "good" component in the market, the function works, otherwise it doesn't. And the fact that the market carefully avoids repetitions and "squeezes" the resulting fluctuations is its nature, and it has been verified by many years of work by hundreds of analysts.



you can turn to algebra :-) What can be said about series Xn=F(Xn-1,...Xn-k); ?
 
Maxim Kuznetsov:
you can turn to algebra :-) What can be said about series Xn=F(Xn-1,...Xn-k); ?
In the simplest cases, it is k-th order.
 
Yuriy Asaulenko:
In the simplest cases, it's MAKs of the kth order.

In the simplest terms, yes. The sum of the weighted values, like TC's in the original post.

My humble recollections of mathematics tell me that such series, depending on the initial vector, will either become periodic-sign-variable or will tend to monotonic function. So extrapolations (predictions, damn it) from them are possible only for these two entities - monotonic growth or strict periodicity