From theory to practice. Part 2 - page 155

New comment
 
Alexander_K2:

In short, the survey showed that the sufferers are not interested in anything other than thinning quotes.

Perhaps, yes - along with market cycles, this thing is an extremely important basis of the Grail.

The various methods of thinning achieve:

1. splitting the original market process into two sub-processes: Alpha and Omega (see https://www.mql5.com/ru/forum/134424)

2. maximum preservation of information, which is carried by tick quotes. Uniform thinning (work with OPEN/CLOSE M1, M5 and higher) leads to the loss of all information and to the transition to a normal distribution of increments, which has the maximum entropy and work with a Wiener process without demolition, which is extremely difficult (but, you can)

3. Adding to the initial series information related to time stamp effects with a certain type of thinning. Even a series (-1; +1) of the "random walk" type can be made more informative by thinning it through an Erlang distribution, obtaining some sort of series for the path (sum of coin increments) through exponential time intervals.

You can check the probability of the next bar up, the probability of the next bar down, the probability of the next bar up, the probability of the next bar down, for example, instead of taking years,
If you want to thin quotes, you can take the same 5000 reports, but thinned, i.e. there will be less of them, calculate the same statistics, make sure it does not work and never come back to this topic. That's it.
But if you have a lot of time, you can spend another ten years "to good use".
 
CHINGIZ MUSTAFAEV:
Instead of worrying about it for years, just check the 5000-minute bars and calculate the probability, for example: upwards bar, upwards bar, downwards bar, downwards bar,
Thin out the quotes however you like, take the same 5000 reports only thinned, i.e. there will be less of them, calculate the same statistics, make sure it doesn't work and never come back to this topic again. That's all.
But if you have a lot of time, you can spend another ten years "profitably".

I have repeatedly done such a trick.

I took thinned tick quotes and minute quotes. In tests, my TS showed much worse results just on М1, all other conditions being equal.

No matter how you slice it, the information is lost on minutes.

 
Quote thinning is a thing of the past. It is time to move on to equity thinning. For example, there is an important unresolved issue in martingale science - how to thin the equity from the inevitable final poker.
 
Aleksey Nikolayev:
Thinning quotes is yesterday's day. It's time to move on to equity thinning. For example, in martingale science, there is an important unresolved question - how to thin the equity from the inevitable final poker.

Loved the post. I like it. I'll give you a big plus.)))

 

Alexander_K2:

Even a series (-1; +1) of the "random walk" type can be made more informative by thinning it through an Erlang distribution, obtaining some kind of series for the path (sum of coin increments) through exponential time intervals.

Would you be so kind as to elegantly demonstrate to us the thinning of said series from all "-1" ? )

 

Not

you can thin out like that and -1 +1 +1 on ticks.

That's a bit of a smile. Try harder to give me the satisfaction of being a poppy)))) on posts like this.

 
secret:

Would you be so kind as to elegantly demonstrate to us the thinning of said row from all "-1" ? )

Squared, perhaps? )

(-1)^2=1^2=1 )

 
Aleksey Nikolayev:

Squared, perhaps? )

We're not looking for easy ways) Erlang suggested there)
 
Alexander_K2:


Ни секунды не сомневаюсь, что ее написал некий Доктор.

Aleksey Nikolayev:

Judging by the sense and goodwill in the text - the article is a translation)

theorists blithely :-)

Quote under discussion:

"Hi, I'm Adam Grimes,..."

<article text>

"Sync & corrections by honeybunny ....

 
Maxim Kuznetsov:

theorists blithely :-)

The quote under discussion:

It's already at the beginning of part 2) Didn't get to it, for I didn't see anything new for the theorists bljat :-)

1...148149150151152153154155156157158159160161162...179
New comment