Distribution of price increments - page 3

 
Maxim Romanov:

The extrapolation cannot be used to predict for several reasons. Firstly, the sampling rate must be correct. If you sample a sine wave with a random time, even a sine wave cannot be predicted. And secondly, how it can be predicted, if it is not known, when will each participant close the order? It is known that all participants who have opened deals will close them, but when, it is already unclear. Or do you disagree with the fact that all open trades will be closed?


Why not? Of course I agree, the deal will be closed at some point. I told you I'm not disputing your point of view. I was only expressing my opinion on the subject.

 
Alexander_K:

The sample is over a million ticks of data.

Take a million ticks, but only the morning ticks.

 

Better yet, take stock exchange quotations.

 

For example, I am also posting tick data for EURUSD in Excel format.

In the attached table:

1. EURUSD" tab, column A - tick data of Ask price

2. EURUSD" tab, column B - tick data of the Bid price

3. EURUSD" tab, column C - price incrementsAsk

4. EURUSD" tab, column D - the Bid price increments

5. Tab "Ask" - histogram of the currency pair quotes

Tab "Returns_Ask" - histogram of price incre ments of the currency pair

For the Bid price everyone can drawthe histograms themselves

Once again we see the non-standardized Student's t-distribution for the price movements (see for example https://en.wikipedia.org/wiki/Student%27s_t-distribution#Non-standardized chapterNon-standardizedStudent's t-distribution) with 2 degrees of freedom and the scaling factors=1.647

I think I won't reveal a secret if I say that Lyapunov's TSP is not fulfilled and the histogram inthe "Ask" tab proves it very well.

Actually, what I, maybe naively, expect from this thread:

Unstandardised Student's distributions are a relatively new section of probability theory and mathematical statistics. And suddenly some really brilliant person will appear on this forum and say something like: "Well, gentlemen, it's clear even to a schoolboy that with increase of a sample of random data the time increments of which obey t2-distribution, distribution of general population tends to xy-square distribution, or Gumbel distribution" :) :) :)

Such a man, indeed, could have been given a monument in his lifetime.

Student's t-distribution - Wikipedia
Student's t-distribution - Wikipedia
  • en.wikipedia.org
Student's t Parameters Support PDF Γ ( ν + 1 2 ) ν π Γ ( ν 2 ) ( 1 + x 2 ν ) − ν + 1 2 {\displaystyle \textstyle {\frac {\Gamma \left({\frac {\nu +1}{2}}\right)}{{\sqrt {\nu \pi }}\,\Gamma \left({\frac {\nu }{2}}\right)}}\left(1+{\frac {x^{2}}{\nu }}\right)^{-{\frac...
Files:
 
Maxim Dmitrievsky:

https://cyberleninka.ru/search?q=garch


Thanks for the links! Interesting enough.

It seems that the creators of these GARCH models are very close to the solution.

Do people using the GARCH models have bad results? And where can we read these trading results? I can suppose that their situation is bad - they don't know exactly for which currency pair a t-distribution must be used, because although for all currency pairs a t2-distribution must be used, each pair has its own coefficient of scale, different from 1. This adds complexity to calculations.

Sincerely,

Alexander_K

 
Alexander_K:

I'm also posting tick data for EURUSD in Excel format as an example.

...

3. Tab "EURUSD" column C - priceAsk increments

...

Tab "Returns_Ask" - histogram of price increments of the currency pair

...

Thanks for the table, it removes many questions. However, others have appeared.

I would like to know what physical meaning you give to the zero increments of Ask rate. So far, from the table presented, I conclude that the non-existent Ask increments are put in when increments occur in another row, in Bid. Is this true?

 
Alexander_K:

What does knowing this distribution actually do? How does it help in Forex trading?

It can help in options trading or risk management. For directional trading, distributions provide nothing. Why everyone discusses them, I wonder too.)

To profit, you need to find patterns, i.e. the dependence of price movement on something, such as a previous movement. In your mathematical terms :) these seem to be conditional distributions. But the "movement" is not necessarily the increments. It's not like you're trading increments.

p.s. You don't have to be a genius here, it's just a research paper on data analysis. The main thing is to look at the essence of the phenomenon, and do not try to mindlessly draw on your previous background.

 
Vladimir:

Thank you for the table, it removes a lot of questions. However, others have come up.

I would like to know what physical meaning you give to the zero Ask increments. So far, from the table provided, I conclude that the non-existent Ask increments are put in when increments occur in another row, in Bid. Is this the case?


You are welcome. I can provide many such tables - for any currency pairs, the quotes of which are provided by the broker.

I interpret the zero increments on a mundane level - as if the buy price was increased at the currency exchange, and the sell price was left the same as it was the previous day.

 
Alexander_K:


The creators of these GARCH models seem to be very close to the solution.

Somehow you didn't notice that they are supposed to model three types of information in increments:

  • trend-uses ARIMA, which uses integer differentiation, or AFRIMA, which uses fractional differentiation (for series with long memory)
  • Volatility - these are various GARCN models, which take into account the nuances of deviation from the mean. For example, as mentioned above: long followed by long and short followed by short
  • distributions.

For some reason you only discuss the latter part of the model actually used.


Do people using GARCH models really have bad results? Where can I read these trading results?

Search for GARCH forex or currency exchange here and you will get a lot of information, including real trading results comparing different distributions used.

EconPapers
  • econpapers.repec.org
EconPapers provides access to RePEc, the world's largest collection of on-line Economics working papers, journal articles and software. We have: for a total of 2,430,512...
 
СанСаныч Фоменко:

The creators of these GARCH models seem to be very close to the solution.

Somehow you didn't notice that they are supposed to model three types of information in increments:

  • trend-uses ARIMA, which uses integer differentiation, or AFRIMA, which uses fractional differentiation (for series with long memory)
  • Volatility - these are various GARCN models, which take into account the nuances of deviation from the mean. For example, as mentioned above: long followed by long and short followed by short
  • distributions.

For some reason you only discuss the latter part of the model actually used.


Do people using GARCH models really have bad results? Where can I read these trading results?

Search for GARCH forex or currency trading here and you will get a lot of information, including real trading results comparing different distributions used.


Thank you!

I will leave the forum temporarily until I learn the information provided, but I will definitely be back.

Sincerely,

Alexander_K