The Sultonov Regression Model (SRM) - claiming to be a mathematical model of the market. - page 3

New comment
 
orb:

Have you read the article, on Sultanov's model? Is there an ISC there, I'm just not aware of it?) Two points are described by ISC and Residuals.

By the way about stationarity you are wrong, faa muses on cointegration (pair trading is based on it I think, well he will speak for himself, I spoke for myself).


Cointegration and is designed for non-stationary time series.

The co-integration is designed for non-stationary time series, and I think it has MNC, and it is precisely MNC only for series with a normal distribution.

 
Demi:

cointegration and developed for non-stationary time series
OK, you misunderstood me or you misunderstood yourself. You wrote that "faa listen, so normality is not necessary, stationarity is not necessary " And isn't cointegration uses such a notion as stationarity of time series?
 

APOLOGIES FOR THE OFFTOPS.

_________________________________________________________________________________________________________________________________________________________________________________________

 

Yes, yes, we got distracted...

Go on, Yusuf! The jackals bark, the caravan comes.

 

Now ask RMS to recognise the parabola Y = a+ bx^2 and it copes with this too perfectly at a=0 and b=1 with an error of 4.78013E-07:

ׂ

ׂ

When a=10000 and b=10:



:

 
yosuf:

Dear forum members, it is no secret that the question of finding the dependencies that describe the basic patterns of the market is an important one. Here we will try to approach this question by all available means of analysis, including various proposals of the participants on this matter and the theoretical and practical material accumulated by this moment from all possible sources. As a result of this work, if we dwell even on just a view of this function, I think we will consider that time and effort have not been spent in vain.

I will begin by demonstrating the possibilities of RMS by using simple examples to describe well-known patterns: linear, parabola, hyperbola, exponent, sine, cosine, tangent, cotangent and others, as well as their combination, which are certainly present in the market. Please support me in this impulse with constructive suggestions and healthy criticism if required.

let's take a price series. let's describe it with a polynomial, a neural network or fourier. we obtain a model that describes this series with almost any accuracy. but this model will never be predictive for the next bar, the same tails. probably better to build a market condition that defines the trend and flat in the early stages of their conception. although there are also many market conditions, but if approached from the profit side this set is likely to be limited to 5 - 10 conditions.
 

Now let's see how RMS "transforms" into a perfect hyperbola Y = b/x at arbitrary b=10 with a fantastically small error of 6.34693E-14 %:

ׂ

:

 
Now show the magic equity parabola on the real.
 
Come on, don't keep me in suspense, when is it going to turn into an idyllic price, or is it going to be a positive balance sheet curve?
 
ivandurak:
Let's take a price series. Let's describe it using a polynomial, a neural network or Fourier. we'll get a model that describes this series with almost any accuracy. But this model will never be predictive for the next bar, the same tails. probably it would be better to build a market condition model that determines the trend and flat on the early stages of their creation. although there are also many market conditions, but if we look at it from the perspective of profit this set will most likely be limited to 5 - 10 conditions.
We will get to Fourier, wait.
12345678910...47
New comment