From theory to practice - page 4

 

A selection of literature on the topic, add to it if you can.

Files:
 


Alexander_K, please tell me if you and the VisSim program system you actively use take into account the inapplicability of the classical probability theory to quotes that is described on p. 4 of Osminin's thesis: "Probability is a very serious problem. 4 of Osminin's thesis:

"Whereas in the stationary case there is demonstrable confidence in the asymptotic consistency of estimates of a particular statistic, in the non-stationary case there is no concept of the general population itself, whichmakes all the developed apparatus of modern mathematical statistics inapplicable, except where the a priori functional identity of the process model is given."

I get the impression that you gravitate towards a single identified type of probability distribution (one of the classical ones, Student's). Is there any methodological error inherent in this?

Greetings Vladimir, I'm already looking forward to your comments - one can feel the approach of a real engineer.

Yes, I must admit - I am not entirely sure about the specific analytical formula for probability density, because it is given for continuous distributions, while we have a discrete one. I do use tabular data in my calculations. But it takes a lot of time to compile tables - that's what I'm doing now. My task was set by myself - TS must deal with all currency pairs at a time. My broker has 36 of them. Now I have processed data for 18 pairs only. I think I'll have enough time before the New Year.

 
Yury Kirillov:

I would also add: "And to the once identified type of average WMA", "And to the once identified method of sampling a sequence of samples"...

As for WMA, I'll defend my point of view to the end. That's exactly the calculation algorithm that corresponds to the very notion of expectation for sampled values.

The method of tick data sampling is not an easy task for me - it's not a trivial one either. I can't believe how many people are working on this problem! But I need a specific, clear and easily transferable algorithm for receiving quotes from one data provider to another. I have found it for myself and theoretically justified it.

 
Yury Kirillov:

A selection of literature on the topic, add to it if you can.

 
Alexander_K:

Thank you, good stuff.

 

I had a similar idea once - to collect a small tick sample and use Student's coefficients to build current market statistics, though without any evolution forecasting, i.e. without using Fokker-Planck and other high mathematics. At that time I thought about an optimal sample size, but never reached the practical constructions - I understood the main problem in time, and therefore gave up this direction. The main problem of this approach, in my opinion, is the following - too small difference of static characteristics of the market, with its different types. Suppose, for example, that 60 ticks are received in a minute, and 60*60=3600 per hour; for a flat market half up, half down; for a trend market the number of ticks in one direction will slightly exceed their number in the other, but this excess will be very small. For example, if 1750/1850 ticks (4 figures), then on the chart it will be a powerful trend - 100 points (4 figures) per hour, and in the sample it will be very difficult to diagnose quickly. Actually this is also described in the preprint the link to which was given by Yury Kirillov .The authors there have analyzed some minutiae increments of EUR/USD pair and as far as I understood they came to similar conclusions.

I'm afraid, but in my opinion, any indicator from the local code base that draws support and resistance levels will work more effectively than your drawings. But in any case, good luck with your research.

 

I would still like to know your opinion on the quote above:

Alexander_K, скажите, пожалуйста, Вы и активно используемая Вами программная система VisSim учитываете ли неприменимость классической теории вероятности к котировкам, отмеченную на стр. 4 диссертации Осминина:

"Если  в  стационарном  случае  есть  доказательная  уверенность  в асимптотической  состоятельности  оценок  той  или  иной  статистики,  
то  в нестационарном случае отсутствует само понятие генеральной совокупности, что делает неприменимым  весь  развитый  аппарат  современной математической  статистики,  
кроме  тех  случаев,  когда  априори  задана функциональная принадлежность модели процесса."

Stationarity and non-stationarity of the market is the basis for all subsequent reasoning.

Or are you proving that all your reasoning applies to a non-stationary market and then it has value, otherwise it is just another bicycle of someone who understands nothing about financial markets at all.

 
СанСаныч Фоменко:

I would still like to know your opinion on the quote above:

Stationarity and non-stationarity of the market is the basis for all subsequent reasoning.

Or are you proving that all your reasoning is applicable to non-stationary market and then it has value, otherwise it is another bicycle of a person who understands nothing about financial markets.


By the way, you used to have a thread on this subject https://www.mql5.com/ru/forum/218475/page17

You abandoned it, although the same question about stationarity and non-stationarity arose there.

Распределение ценовых приращений
Распределение ценовых приращений
  • 2017.11.12
  • www.mql5.com
Уважаемые трейдеры...
 

actually, you have to ask practicing traders: they may not know statistics, but they understand how the market works, and they build the same distribution of volumes, on the borders and centres of these distributions they build levels, fair price zones and so on... to be honest - it works, but also honest - sometimes, because the market jumps from state to state and it's hard to catch them... especially on ticks - I don't understand what you can find there and what size signals will be, a couple of points?

 
Maxim Dmitrievsky:

In fact, you have to ask practicing traders: they may not know statistics, but they understand how the market works, and they build similar volume distributions, on the borders and centres of these distributions they build levels, fair price zones and so on... to be honest - it works, but also honest - sometimes, because the market jumps from state to state and it's hard to catch them... especially on ticks - I don't understand what you can find there and how big the signals are, a couple of points?


A couple of pips on ticks is millions on the daily market. This is what attracts.