Market phenomena - page 8
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
Pakukas, the phenomenon is there, the reasons for not seeing it are very simple:
but what if it's there? Paint your beard a pungent yellow? All right, just repaint the beard on your avatar. And if it's not there - I'll leave the forum and I certainly won't bother with your inability to use TA and VA anymore. Deal? :о)))
Pakukas, the phenomenon is there, the reasons for not seeing it are very simple:
but what if it's there? Paint your beard a pungent yellow? All right, just repaint the beard on your avatar. And if it's not there - I'll leave the forum and I certainly won't bother with your inability to use TA and VA anymore. Deal? :о)))
Not fair. The forum will suffer. A loss, a substantial one. "... If it's not there...", your avatar will have a beard.
Oh, my God, you're so aggressive. I told you, it'll turn up. We have to wait.
In any case, don't agree to paint your avatar. It's another scam. Like, "at least the avatar."
You can dye yours, it will grow black in a couple of months, but the avatar won't grow a thing. Later on it will also be called a phenomenon and another nail in the coffin of TA.
The phenomenon I want to post may or may not be known to anyone, or may not be known to everyone. Anyway, I haven't registered it anywhere. Let's take EURUSD M15 (Alpari data for about 10 years) and see its increments.
Over 10 years Alpari has some of the data is 4 digit (reduced to the 5th digit), some is really 5 digit. Do you have a histogram of increments in 0.0001 or 0.00001 increments?
And for which increments do "dips" appear on the histogram?
Farnsworth:The 5th digit was introduced not so long ago (maybe a year or even less) and generally speaking it doesn't affect the result. You can see it in the dynamics of the alpha and omega processes, if you look at them carefully. The step of the histogram is greater than 0.0001, I can't say exactly now, but the phenomenon appears on the number of sites 500, i.e. roughly speaking Max(Open)-Min(Open) divided by 500. It would hardly even have an effect if the variable was continuous.
PS: "Histograms" are not built by me, but by MathCAD. You might be surprised, I also know how to build them. I don't think you need to look for an error in histogram construction, just check on the data.
Generally speaking such "superpositions of distributions" have occurred to me, and it used to happen just on a mixture of 4 and 5 digit data. The fact that the 5-digit history is much shorter rather than hindering the separation.
That's why for the check it would be worth doing a separate distribution for 4-digit times and 5-digit times, if I'm not mistaken, it's about 2009 in Alpari.
P.S. By the way, hello there :)
This is a very subtle question, by the way. The Puppeteers 2 have done a great job here. Not the ones who are real puppeteers, but the ones who are DCs. Incidentally, this is another phenomenon, but rather a negative one.
When researching distributions, I stopped choosing mono-broad intervals a long time ago. More like dividing into quantiles. But there is a hitch here as well: at some individual values of returns (for example, 0.0004) their concentration is too high to qualitatively select intervals by quantiles.
The sampling error of the data (0.0001) is quite large to affect the quality of the histograms. This is indirectly indicated by Prival, by the way. That is, the distribution, which can formally be considered continuous, is not so much continuous, but very nasty - discrete-continuous.
Example: take the 1H or 4H EURUSD returns on the history since 1999 and try to plot quantiles 0.02, 0.04, ..., 0.98 (50 quantiles) on this set. Excel, of course, will formally do it correctly, but if you recalculate the number of values within each interval, they will differ greatly (although they should nearly coincide). And they will differ not by percent, but sometimes by times!
At first this was very tiring, but then I found a solution: I began to add to returns a deliberately small random value, much smaller than 0.0001. And everything worked: quantiles became similar to real quantiles, i.e. amounts of values falling within each quantile interval now differ by units, i.e. by tenths or hundredths of a percent.
This "handling" has almost no effect on the data, because this data is already distorted by the DC filter by an order of magnitude of the spread.
Generally speaking, such "distribution superpositions" have occurred to me, and have happened just on a mixture of 4 and 5-digit data. The fact that the 5-digit history is much shorter contributes to the separation rather than hinders it.
That's why for the check it would be worth doing a separate distribution for 4-digit times and 5-digit times, if I'm not mistaken, it's about 2009 in Alpari.
P.S. By the way, hi there :)
I will check later, in a couple of days, but having discovered the phenomenon I analysed the plots - everything is stable. Again, this can be seen on the dynamics of the processes themselves:
The fifth sign has no effect, the step is used much more than 0.0001. Well yes, 1-1.5 years t.p.s. seems to have some fluctuation, but I think it is me who has not introduced the classification very qualitatively.
Another phenomenon is long-term memory.
Most of us (of those who do this, of course) are used to measuring market memory by Pearson correlation - more precisely, autocorrelation. It is well known that such correlation is quite short-lived and significant with lags of up to 5-10 bars at most. Hence it is usually concluded that if the market has a memory, it is very short-lived.
However, Pearson correlation is only able to measure linear relationships between bars - and virtually ignores non-linear relationships between them. The correlation theory of random processes is not called linear for nothing.
However, there are statistical criteria that allow us to establish the fact of an arbitrary relationship between random variables. For example, the chi-square criterion - or the criterion of mutual information. I haven't really bothered with the second one, but I have bothered with the first one. I will not explain how to use it: there are plenty of manuals on the Internet, which explain how to use it.
The main question was this: is there a statistical relationship between bars which are far away (for example, if there are a thousand bars between them)? There was no question about how to use it in trading.
The answer is yes, it does exist, and it is very significant.
For example, if we take the EURUSD history from 1999 on H1 and check the chi-square for pair returns, we find out that in the range of "distances" between bars between 10 and 6000, in about 90% of cases the current bar depends on the bars from the past. 90%! At distances between bars of more than 6000 such dependences occur less frequently, but still occur!
Frankly, I was stunned by this "discovery" as it directly shows that the euro has a very long term memory. On EURUSD H1 6000 bars is about a year. This means that among the hourly bars of a year ago, there are still bars that the current zero "remembers".
On H4 significant dependence is found up to about 1000-1500 bars. I.e. the duration of "market memory" is still the same - about a year.
Recall Peters who says that the market memory is about 4 years. Contradiction, however... I do not know yet how to solve it.
Not having calmed down, I decided to check if my chi-square would show such dependencies if I fed the input synthetic returns generated independently. I chose two possible distributions of the synthetic returns - normal and Laplace - and ran it. Yes, it shows, but within the significance level of the criterion (I had 0.01)! In other words, the synthetic showed about 1% dependent bars in the past - just at the level of probability of criterion error.
What are the conclusions?
1. Euro quotes are definitely not a Markov process. In a Markov process the current value depends only on the previous value. In our case we have numerous bars in the very distant past, on which the current bar depends.
2. The so-called "foundation" certainly plays a certain role - let's say, as an excuse to move the quotes. But it is certainly not the only one. We need to look at the technique!
3. This result is still purely theoretical and has no practical importance. Nevertheless, it clearly shows, that not all is lost for those who look for something.
Another phenomenon is long-term memory.
The main question was: Is there a statistical relationship between heavily distant bars (e.g. if there are a thousand bars between them)? There was no question about how to use this in trading.
The answer: yes, there is, and a very significant one.