FR H-Volatility - page 37
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
It seems to me that the trend is a property of the market. If you look at the right edge of the chart, the duration of ZZ increases (the spread on the axis
Y more), it would be strange the other way round if the spread was increasing and there was no trend. I don't understand the sadness about that, though.
Candid, my results are about the same. So it's not a bug in your algorithm. Or it's a bug in mine, too. :)
Yep, got it, thanks. It remains to be seen if it can be used in some way :). Or think over the algorithm of EZ that takes this property into account.
I don't understand the sadness about this, though.
For example there are doubts in correctness of averaging FFT results on long periods.
Then it will become clear that the right edge bulging is the result of a certain market character around the 3000 reference. If you don't mind, post it with such a dummy instead of polynomial regression.
It's not quite clear why the 4th degree polynomial essentially ignored the vicinity of the 3000 reference. In principle, if I want to look for some correct(i.e. predictable) oscillations, I'll apply fft to the data to start with. But for now the trend issue is much more important to me, or rather whether the trend is a property of my algorithm or a property of the market. If you don't mind, post similar data for your favourite zigzag algorithm.
P.S. I would like to point out that short oscillations cannot cancel the trend in any way, only oscillations with a period significantly longer than the interval covered by the graph can cancel it.
Ignored because the 4th degree is not sufficient to even slightly represent the bends (to put it mildly) of your curve. The 4th degree has only two inflection points. How can you show any kind of oscillation? To be quite precise, it (your polynomial) shows that the right maximum is higher than the left one. Well, it can't do much more than that.
There can be neither right nor predictable oscillations on forex. And ffs can be applied to wrong ones as well. I suggested the dummy to underline by elementary means that there are stochastic oscillations. And, since the price process has a fractal structure, these oscillations (as well as tick oscillations - compare the time intervals between ticks) will have stochastic nature at any scale.
I don't mind. Although I won't load 8 years for that. Look for the last two years. I like any ZigZag, that's why I took such, that on a quarter of your period it gives a quarter of your ZZ segments. At the same time, I also took a scale with period N=50, and a linear regression.
The Y-axis is the segment duration in minutes. It was calculated not by subtracting the start time from the end time (in this case, passing through Saturday-Sunday could give an additional 2880 minutes), but by the number of minute bars in the segment. Please note that this is not an exact calculation, as in a sluggish market, minute bars are known to drop off. This means that the actual duration of individual segments may be several units (or even tens) longer. Since you didn't name the instrument I did for EURUSD. My highs get to 4000 and yours only to 3000. This is probably due to the fact that we have different pairs.
Ignored because the 4th degree is insufficient to even marginally represent the bends (to put it mildly)
...
The Y-axis is the duration of the segment in minutes. It was calculated ... by the number of minute bars in the segment.
...
Since you didn't name the instrument I did for EURUSD. My highs reach 4000 and yours only 3000. I guess it has to do with the fact that we have different pairs.
Yeah, thanks.
The 4th degree was mentioned in the context of your suggestion that a single wisp can shift the entire grade, certainly one of those should be able to sense it. However, the 10th degree does not feel anything either.
I was calculating the duration of the segment in the same way.
The pair is basically unimportant, I wrote that this pattern is observed for all majors. And for the divergence with the highs, the answer is simple - just in case, I wrote the statistics separately for the "down" segments and for the "up" segments. The general behaviour turned out to be the same for them, but the graph shown is only for half of the segments. That is, the X-axis in my picture is compressed to half of yours. Or, in other words, it corresponds to twice as many statistics for the same time interval.
Now my conclusions: A small trend is noticeable in your picture too, despite the different (through my fault) X scale and unfortunate (for the effect) phase - the end of the sample falls on a fluctuation "against the trend". That is, the effect is confirmed, although the measure of its usefulness remains uncertain :)
You probably can't use this phenomenon.
The wag was rising by more than 70%.
The figure on the left shows the time of ZZ top formation in black and its amplitude in red. Data on EUR/USD from 2004, plotting step H=10 pips. no strong trends are visible. And what may this trend mean? Realistically, only the time between formation of neighboring tops of ZZ increases. It means that the market becomes less volatile or more antipersistent. The second variant looks more attractive, because this property can be used in TS. The increase of antipersistence should be reflected in the amplitude of GZ segments - they will decrease, as can be seen at the end of the figure.
The red color on the right shows the distribution of the GZ side amplitudes and the black color indicates the time of the next vertex formation. Generally speaking, these two parameters, separately, indicate the area where the next peak is expected to be formed, one on the time scale, the other on the price scale. I.e., we get the coordinates of the target, where the apt strike should be made. This is what Prival told us about for a long time! For example, it follows from the above constructions that the next peak in GZ will most likely be formed in ten minutes, and its amplitude will be about 15 points.
Now my conclusions: A small trend is also visible in your picture, despite the different (my fault) scale of X and the unfortunate (for the effect) phase - the end of the sample falls on a "counter-trend" fluctuation. So the effect is confirmed, although the measure of its usefulness remains uncertain :)
Some time ago in this thread there was a discussion with kamal about volatility. A small excursus in which, among other things, it was said that volatility is persistent. If there is a statistical relationship between the size of a WP segment and its duration (which is easy to check by plotting the points corresponding to the WP segments on the size-time plane and drawing a regression line; all the points only, not separately up or down), then the existence of such oscillations is justified by this fact. Segment size actually represents volatility on some scale, and its relationship to period (if there is one) leads to the fact that an increase in size also leads to an increase in duration. If the volatility is persistent, such an increase will have a meaningful lifetime, i.e. it will form a local trend.
The figure on the left shows the time of ZZ top formation in black and its amplitude in red. Data on EUR/USD from 2004, plotting step H=10 pips. no strong trends are visible. And what may this trend mean? Realistically, only the time between formation of neighboring tops of ZZ increases. It means that the market becomes less volatile or more antipersistent. The second variant looks more attractive, as this property can be used in TS. An increase in antipersistence should be reflected in the amplitude of ZZ segments - they will decrease, which can be observed at the end of the figure.
Hi Sergey !
I disagree with the highlighted phrase in your post. This is a pattern I discovered a couple of years ago: there is a statistical linear relationship between segment size and duration. So when there is uptrend in segment duration, it is not a decrease but an increase in volatility. Admittedly, this relationship is statistical and I have not found any change in the shape of SP when volatility changes. So I do not know how it can be used.
Generally, these two parameters, separately, indicate the area where the next top is expected to form, one on the timeline and the other on the price scale. In other words, we get the coordinates of the target, where to strike. This is what Prival has been talking about for so long!
Alas, it's not that simple. It's not even about the width of the distributions. We can construct a confidence interval in which, say, 90% of reversals occur. I have done such constructions.) It appears that the price may stay in this range for a long time, which means that the problem is the prediction of the moment when the price leaves this confidence interval.
Yura, hi!
Let me disagree with you too. Fig. below shows the dependence of the time of formation of the next WP vertex on its amplitude (red dots).
The black dots show a linear least-squares regression plotted for the case of nonuniform partitioning along the abscissa axis. As you can see, we can talk about a very weak time dependence on the amplitude, i.e. there is practically no dependence - we are talking about a time gain of 4 minutes when the amplitude changes from 10 to 100 points.