From theory to practice - page 204

 

Still very old, but I think interesting information regarding the receipt of ticks before processing.

https://www.mql5.com/ru/forum/102066/page9

Стандартные заблуждения в попытках торговать в шуме (было "Кошмар на улице МТ4")
Стандартные заблуждения в попытках торговать в шуме (было "Кошмар на улице МТ4")
  • 2006.12.18
  • www.mql5.com
Информация - самый ценный продукт, который может быть... Для форекса корректная база данных решает всё...
 

With this post I bow low to the mathematicians.

Look at the probability distributions of time intervals (in seconds) between real tick quotes.

For the AUDCAD pair:


For the AUDCHF pair:


I dare say that when the number of accepted ticks increases to 1,000,000, the values of the probability density function (the "Probability" column) will be almost the same.

Hypothesis - we have just the forex market time scale in front of us. Work (receive data, make calculations) must be performed on this time scale, not on the uniform or exponential one.

Please help me determine the analytical formula for the probability density function of this distribution! And as a consequence - the formula for the generator of random numbers of this distribution.

Regards,

Alexander_K

 

The distribution of incremental velocities:

I have no idea what it is. But, interesting.

 
Alexander, can you send me the raw data and I'll try to find an allocation?
 
Alexander_K2:

The distribution of incremental velocities:

I have no idea what it is. But, interesting.

It looks like a two-sided exponential with discontinuity noise.

Laplace distribution

 
Alexander_K2:

...

Please help me determine the analytical formula for the probability density function of this distribution! And as a consequence, the formula for the random number generator of this distribution.

Regards,

Alexander_K

Simulation of random variable with specified distribution law

 
Dennis Kirichenko:
Alexander, please send me the raw data and I will try to find a distribution.

Hi Denis.

The columns are named on Sheet 2 to make it clearer.

Files:
AUDCAD.zip  5512 kb
 
Alexander_K2:

The distribution of incremental velocities:

I have no idea what it is. But, interesting.

what are the minus velocities in your diagram there?
 
Nikolay Demko:

It looks like a two-sided exponential with discontinuity noise.

Laplace distribution.

Nope, no way - it's not the Laplace distribution, nor is it a one-sided geometric distribution.

I thought it was this one:

with the formula:

but, it didn't fit...

 
Alexander_K2:

With this post I bow low to the mathematicians.

Look at the probability distributions of time intervals (in seconds) between real tick quotes.

For the AUDCAD pair:


For the AUDCHF pair:


I dare say that when the number of accepted ticks increases to 1,000,000, the values of the probability density function (the "Probability" column) will be almost the same.

Hypothesis - we have just the forex market time scale in front of us. Work (receive data, make calculations) must be performed on this timescale, and not on the uniform or exponential one.

Please help me determine the analytical formula for the probability density function of this distribution! And as a consequence - the formula for the generator of random numbers of this distribution.

Regards,

Alexander_K

There is a hypothesis: "there are fewer ticks at night and more ticks during the day, so the price can travel more in a minute during the day and less at night" which is true if we assume that the price always travels approximately the same distance per tick.

You'd better measure the average distance a price travels per tick for different times of day.