Discussion of article "Evaluating the ability of Fractal index and Hurst exponent to predict financial time series"

 

New article Evaluating the ability of Fractal index and Hurst exponent to predict financial time series has been published:

Studies related to search for the fractal behavior of financial data suggest that behind the seemingly chaotic behavior of economic time series there are hidden stable mechanisms of participants' collective behavior. These mechanisms can lead to the emergence of price dynamics on the exchange, which can define and describe specific properties of price series. When applied to trading, one could benefit from the indicators which can efficiently and reliably estimate the fractal parameters in the scale and time frame, which are relevant in practice.

Demonstration of indicator operation on real data

The indicator is requested to evaluate 600 days with the evaluation window of 64 points. The result contains 536 values of the fractal index and it is shown in Fig.6.

FigGAZP

Fig.6 Close prices of Gazprom and the fractal index evaluation results

The figure shows the correlation of the index values and the behavior of prices. The blue color of the index graph corresponds to the trend state of the system, it indicates the trend stability and the ability to predict future behavior. Violet color indicates anti-persistence of the "pink noise" type, which corresponds to "negative" memory and flat. Yellow corresponds to the "Brownian motion", i.e. the movement is random and cannot be predicted.

Author: Roman Korotchenko

 

I read the article, but I did not see what I expected when reading the title of the article: where is the possibility of predicting financial BP ? - it is not there or is it there ?

 
Where does the SSA get it from?
 
Dmitriy Skub:
Where do I get SSA from?

I don't think the article mentions SSA, is it needed in the sources?

If yes, then probably you should go to the author in the market, I asked him last year https://www.mql5.com/ru/forum/190847/page3#comment_8183848.

Обсуждение статьи "Прогнозирование рыночных движений с помощью байес-классификации и индикаторов на основе сингулярного спектрального анализа"
Обсуждение статьи "Прогнозирование рыночных движений с помощью байес-классификации и индикаторов на основе сингулярного спектрального анализа"
  • 2017.04.28
  • www.mql5.com
Опубликована статья Прогнозирование рыночных движений с помощью байес-классификации и индикаторов на основе сингулярного спектрального анализа: Авт...
 
Igor Makanu:

I read the article, but I did not see what I expected when reading the title of the article: where is the possibility of predicting financial BP ? - is it not there or is it there ?

Predictability, it is there.

 
Igor Makanu:

The article doesn't seem to mention SSA, is it needed in the sources?

If yes, then probably in the market you need to go to the author, I asked him last year https://www.mql5.com/ru/forum/190847/page3#comment_8183848.

So I'm asking the author)))

There is an includnik used from the SSL directory - in the attached archive.

 
Igor Makanu:

I read the article, but I did not see what I expected when reading the title of the article: where is the possibility of predicting financial VR? - is it not there or is it there?

It is there. Only not prices, but entry points.
 
And most importantly, fractals are the basis of the elementary structure of market price dynamics (unlike all other figures, it is constant (present constantly on the chart), consistent and unchanged in shape (only the shape parameters change). The structure is identified and justified within the framework of the impulse equilibrium theory.
 

Didn't understand what the article was about. Maybe there will be a sequel...

"Fractality" is not measured in any way by the Hurst index. It can be used to assess the persistence/antipersistence of BP, no more, no less. There is no "pink noise" and "negative memory" in the market, of course.

Any Hurst index will describe Brownian motion, not only in yellow colour (fractional at H != 0.5) and will in no way suggest the presence or absence of market "memory", except in terms of long-term dependencies

Moreover, talking about "memory" in a 64 point window is, generally speaking, not serious. A process either has memory by itself (and it is long-range, for the whole depth of the process) or it has no memory at all. Initially, Hurst and R\S were interpreted correctly, as long-range dependence in the form of a trend component. But as the window gets smaller, it loses any common sense at all (the smaller it is, the less sense this indicator makes)

 

how correct is it to evaluate financial BP for fractality using only one Close price ?

SZY: I read an article on hubra about calculating fractal dimension using box-counting dimension, in my opinion it is more correct, although maybe I liked the style of presentation of the material on hubra better ))))

Вычисление фрактальной размерности Минковского для плоского изображения
Вычисление фрактальной размерности Минковского для плоского изображения
  • habr.com
Доброго времени суток читатель. Сегодняшний пост будет посвящен вычислению приближенного значения фрактальной размерности плоского изображения, которая тесно связано с размерности Минковского. Это интересно как минимум по двум причинам. Во-первых оказывается, что размерность ограниченного множества в метрическом пространстве может быть не...
 
Igor Makanu:

The article doesn't seem to mention SSA, is it needed in the sources?

If yes, then probably you should go to the author in the market, I asked him last year https://www.mql5.com/ru/forum/190847/page3#comment_8183848.

For the indicator in this article SSA is not needed. All codes are attached to the text.