Machine learning in trading: theory, models, practice and algo-trading - page 1351
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
Renko
based on it these data are collected in increments of 0.001Yep....
There are the highest level specialists here ))))
.....
Renco assumes a fixed step on each point in the data
which means the distribution of a given reenco series would look like this
Mda....
There are the highest level experts here ))))
.....
Renco assumes a fixed step on each point in the data
which means the distribution of the predifferentiated renko series would look something like this
You're confused. The delta high catch is always not more than 0.001 clause equals high and low depending on the rise and fall. there's nothing there
Specs are not specs, but there is no grail, neither me nor you. And everyone can complicate thingsWhere did this figure come from???, there are 6 columns in the file, the first of them is time, so it will not be taken into account, the remaining columns are 2 to 6
Let's build the distributions for all the columns .....
The first 5 distributions are based on the original rows, the second 5 (the ones below) are based on the distributions of the rows
So where did these drawings come from?
Let's take, for example, column E - price CLOSE
............
The increments, respectively, are:
.......
So, the histogram of these increments has this strange appearance. And it's not Renko!
If this is your first time working with increments, just try to predict the "+" or "-" sign of the next increment, like Doc did, please.
Gentlemen!
Grail is needed like air, so I have to go to extreme measures, but without breaking confidentiality of information.
Let us say: I have in my PM the data on the basis of which a certain Mr. X received the coveted Grail (see attached archive).
The format of the data is standard: Time, OHLC, Volume.
The method of obtaining these data is non-standard and unique. The CLOSE increments, for example, have the following distribution density:
I can't get my hands on it...
Please, who isn't busy - see if you can extract unrestrained profits from these rows using a forest/neural network.
Especially those who work with returns - I'm begging you.
Need a simple answer: the Grail is there/Grail is not there.
Thank you.
What kind of story is so tattered? Give a normal one, without any omissions, then maybe it will make sense to look something...
In addition to the posthttps://www.mql5.com/ru/forum/86386/page1344#comment_10678027 about predicting random processes.
Now it's the same, we use a neural network to predict the value of LF filter output on real market data. Prediction interval - 5 min on 1 m TF.
All values are in conventional units on the scale of NS data, of course, may be recalculated into real values.
With prediction values in 5 m < -(3-4) or > (3-4) you may work in a real TS. In the opposite direction will not go.
Well, and by the way, forecast at 10 m
Recall that if forecast time = 0, we get the straight line at an angle of 45 degrees.
Gentlemen!
Grail is needed like air, so I have to go to extreme measures, but without breaking confidentiality of information.
Let us put it this way: I have in my PM the data on the basis of which a certain Mr. X received the coveted Grail (see attached archive).
The format of the data is standard: Time, OHLC, Volume.
The method of obtaining these data is non-standard and unique. The CLOSE increments, for example, have the following distribution density:
I can't get my hands on it...
Please, who isn't busy - see if you can extract unrestrained profits from these rows using a forest/neural network.
Especially those who work with returns - I'm begging you.
Need a simple answer: the Grail is there/Grail is not there.
Thank you.
What kind of story is so tattered? Give me a normal one, without gaps, then maybe it will make sense to look at something...
I gave you the one they sent me. This man never contacted me again. This is not a fairy tale - please believe me.
I am VERY interested in this side-by-side - have never seen these before.
If anything turns up - we'll figure out the algorithm for its formation together. I have only sketchy information about it - obviously, this person didn't really want to talk about it :))
I gave the one they sent me. The man never contacted me again. This is not a fairy tale - please believe me.
I'm VERY interested in this guy - I've never seen anything like this before.
If we get anything at all, we'll figure out the algorithm for its formation together. I only have scrappy data about it - obviously, this person didn't really want to talk about it :))
Let's assume that these are decision points. Then we need to determine what kind of tool it is and restore the history for generating predictors, because he obviously had a history, not just raw numbers with an incomprehensible discretization.
There will be a history, you can think about what these points are, overlaying them on the graph.Do you know how to make money on such a distribution?
If I knew, I would not have asked for help here.
In general, the story with the appearance of this file in my possession is unclear. I do not understand the motives and goals of the person who sent it. But it would be very foolish not to investigate it. IMHO.