Recognising images ( rhetorical theme ) - page 4
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
Good afternoon...
It's an interesting topic in itself, of course any image can be digitised
then by averaging the image... for example to distinguish between a cat or a dog in the picture :-)
The task is very interesting indeed. I have been working on it for nearly two years and have got good results, but I haven't got my Expert Advisor for real yet (personal qualities are to blame :).
IMHO, any image (figure) consists of points connected by segments. So the problem to recognize the image, fades in front of a successful algorithm that can split the movement of these very segments, in a simple zigzag.
Just like many others have tried to look for patterns based on ZZ, but nothing good came of it. But if we add the FZR filter to this zigzag, the results will be better but it is not the limit.
I have now manually developed a method of breaking up the chart into waves without using waves and ZZ based on time. I am waiting for great results ;)
In general, I see the flag on the horizon - "fractal analysis" and it cannot be done without analysis of images and different periods!
The task is really interesting. I have been developing it for almost two years, I got good results, but I have not got my real EA yet (my personal kachestva is to blame :).
IMHO, any image (figure) consists of points connected by segments. So the problem of image recognition, pales before finding a successful algorithm that is able to break the motion in these very segments, in simple zigzag.
Just like many others have tried to look for patterns based on ZZ, but nothing good came of it. But if we add the FZR filter (fractal-zigzag filter ) to this zigzag, the results will be better, but it is not the limit.
I have now manually developed a method of breaking up the chart into waves without using waves and ZZ based on time. I am waiting for great results ;)
In general, I see the flag on the horizon - "fractal analysis", which cannot be done without image analysis!
What kind of beast is that in your mind?
When I was young, I read Masterforex, and that's where the definition came from. Briefly, it means that one of two neighbouring peaks must be outside the channel constructed by two preceding peaks. So ;-)
When I was young, I read MasterForex, and that's where the definition came from. In short, it means that one of the two neighbouring peaks must necessarily be outside the channel built by the two peaks preceding it.
Red, assume a zigzag, green line on top, is the same as red, but the peaks are filtered by the FZR rule, i.e. if a new peak appeared on a RE satisfies the FZR rule, then it is drawn on the modified RE.
So we are analysing the last 4 peaks? Or troughs?
FZD (Fractal Zigzag Pivot)
The notion of FZD was introduced by Masterforex-V. There are two types of FZD: reversal and correctional. Both types are shown in the figure below. To understand the concept of a PDR, you should note that a PDR changes from support to resistance on a downward move and vice versa, from resistance to support on an upward move. This fundamental and fundamentally important phenomenon of changing levels always occurs when the trend changes direction. In the case of a reversal FZR, the base of the 5th wave of the previous trend is broken and the price fixes below this level, after which the movement begins in a new trend. In practice, it is necessary to draw a horizontal line at the base of wave 5 on the M5 and M15 charts and expect a breakout and fixation behind it. In a corrective FZR, formation of 1 and 2 waves in the opposite direction from the previous movement, and then a breakout of the top of the 1 wave and consolidation behind it takes place. In this case, we draw a horizontal line along the top of wave 1 and wait for the price to break through this level and consolidate behind it. No trend change can occur without this change in levels. The duration of the upcoming movement can be predicted based on the time frame in which the FTR occurred: M5 and M15 - a short-term trend, H1 and H4 - a medium-term trend. There is one more factor that confirms the FZR. These are moving averages. Their relative position and the price position in relation to the MA either confirm or cancel the FZR. In a true FZR, the price is pulled over the line of level change (the base of the 5th wave or the top of the 1st wave). Breaking and fixing the price behind the level change line on the M5 timeframe determines the daily trend.
Thanks, haven't got a clue yet, but I'll look it up on the internet. Seems a very complicated definition of a simple thing
Will post the turkey pictured in the screenshot later in this post
The task is really interesting. I've been developing it for almost two years and have got decent results, but I haven't got an EA for real life yet (personal qualities are to blame :).
IMHO, any image (figure) consists of points connected by segments. So the problem to recognize the image, fades in front of a successful algorithm that can split the movement of these very segments, in a simple zigzag.
Just like many others have tried to look for patterns based on ZZ, but nothing good came of it. But if we add the FZR filter to this zigzag, the results will be better but it is not the limit.
I have now manually developed a method of breaking up the chart into waves without using waves and ZZ based on time. I am waiting for great results ;)
In general, I see the flag on the horizon - "fractal analysis" and it cannot be done without analysis of images and different periods!
Also worked on such a task 4 years ago - was my first development for forex trading. Price patterns are really easy to recognise, especially with a zig-zag. Recognizing a cat from a dog is a much harder task. In my experiments EAs based on pattern recognition have been losing. Even if you forget about double tops and bottoms, flags, triangles and other that geometry and just look for the most similar parts of history (nearest neighbour method), you still get a loss. Do people really pay for geometry in forex? What a bunch of weirdos!