a trading strategy based on Elliott Wave Theory - page 234
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
Well, I'm not arguing, I'm participating in the discussion. :-))
Besides you speak as a statistician, and your arguments are statistical.
I, for example, find it easier to understand the nature of a process, than its statistics. The same applies to
indicators. So about N-Hurst I agree with you quite well. Nevertheless, I look forward with interest
I look forward to seeing what you come up with when you get your act together.
Please comment on the observation.
Sergey, I've already written earlier that I modified the correlation function to my task, by the way, you also like to modify something. Let me remind the point of modification. I was interested in the strength of correlation between samples in its qualitative physical interpretation: where the strength of correlation should be measured from 0.0 (no strength) to some maximum value for a particular series. For prediction, I use:
(1) the values themselves
(2)the shape of the curve, or rather its local extrema.
I should also remind you that you can see the autocorrelation function (statistics) represented in the charts, but I still chose equality to zero as the criterion, and it is not quite correct. The criterion F itself should have the form [0;F] and should be calculated depending on the specific characteristics of the series. I am still thinking and experimenting.
A Wiener process
Now about the Wiener process. The criterion for it doesn't have to be zero!!!! Just look closely at your formula for this process. In this case, it's a short memory process, not a total lack of memory for that process at all. What makes you think that? And you can't make money from these processes just because that memory is very "short".
Noise
And to be absolutely sure that it works, just consider noise in its pure form. Here is an example of its generation:
For noise, the strength of the connection between elements is ALWAYS zero and the graph always looks like this, there are no options at all:
So, the modification itself is not an error, everything works perfectly correctly.
A nosy mind must have noticed the difference in the strength of the connection between the counts for the EURUSD ticks and the minutes series...
A nosy mind has noticed that the correlation function graphs are represented in log coordinates. This is the first time I've seen log coordinates used for a correlation function. Comment on the meaning of their use. And I would like to see the graph itself in normal coordinates.
Well I'm not arguing, I'm participating in discussion. :-))
You speak as a statistician, and your arguments are statistical.
I, for example, find it easier to understand the nature of a process, than its statistics. The same applies to
indicators. So about N-Hurst I agree with you quite well. Nevertheless, I look forward with interest
I look forward to seeing what you come up with when you get your act together.
By the way, I would also like to say something on the subject. I'm not arguing either,
just exchanging opinions. If there's anything that seems harsh
I'm not, it's just a matter of penmanship.
My criterion, having processed this text, shows a strong correlation between the letters and a covert allusion to "drink beer" :o)))) I wonder if the criterion is lying or not, or am I just in a good mood?
Если вдруг где то покажется резкость в моих словах, это не так, это издержки общения по переписке.
My criterion, having processed this text, shows a strong correlation between the letters and a hidden allusion to "drink beer" :o)))) I wonder if the criterion is lying or not, or am I just in a good mood?
This question cannot be answered statistically reliably.
The fact is that the word 'beer' is implicitly present in 98.5% of the words used.
But, unfortunately, the word 'drink' is completely absent. So, Sergei, keep working on the criterion.
To fulfil my terrible oath, I am posting the autocorrelation function for EURUSD, M1, 2006.
If you, Sergey, would tell me how to save a picture of the working area
or at least its graphics without a headache, life would be much more cheerful.
There's no doubt that displaying static results in Matkad is much more enjoyable.
For research involving calculations and their graphical representation, it's a wonderful thing.
However, working with graphical representation of large arrays is hardly possible.
Anyway, I can now handle the "dirtiest" part of your calculations. :-))
Thanks for the science.
Если вдруг где то покажется резкость в моих словах, это не так, это издержки общения по переписке.
Мой критерий, обработав этот текс, показывает сильную связь между буквами и скрытым намеком на «выпить пиво» :о)))) Интересно, критерий врет или нет, или у меня просто хорошее настроение??
This question cannot be answered statistically reliably.
The fact is that the word 'beer' is implicitly present in 98.5% of the words used.
But, unfortunately, the word 'drink' is completely absent. So, Sergei, keep working on the criterion.
:о))))
It's strange that the word "beer" is found without the word "drink". I would even say suspiciously...
Now about the Wiener process. The criterion and it shouldn't be zero for it!!!! For that, just look closely at your formula for that process. In this case, it is a process with short memory, not at all with no memory at all for that process. What makes you think that? And you can't make money on these processes just because this memory is very "short".
Sergey, FAC is built for FIRST DIFFERENCES. Look at the formula and take the differences between adjacent terms, you will see that in the end, only sigma is left - a random variable. Therefore, the FAC for a Wiener process is identical (to within 1/SQRT(n)) to zero on any TF and between any counts. This is strictly mathematically proved. And one cannot make money on these processes only because there is NO MEMORY at all!
What's all this pathetics? Do you think this is a better way to present the results
than this?
It's just that I try to make the picture as informative as possible. Logarithmic scale on the horizontal axis in this case, it seems to me, contributes to it.
to Yurixx
If you had told me to save a picture of the workspace or at least a graphic, my life would have been much more exciting.
There's no doubt that displaying static results in Matkad is much more enjoyable.
For research involving calculations and their graphical representation, it's a wonderful thing.
However, working with graphical representation of large arrays is hardly feasible.
Anyway, I can now handle the "dirtiest" part of your calculations. :-))
Thanks for the science
I do not know how to save a picture. I use a screenshot myself :-(.
As for displaying "significant" data arrays, a sensible approach is important here. 10^6 pixels is more than enough. If the vector is longer, nothing prevents to display it through a point or two, then by induction. The eye will not notice it.
By the way, Yuri, did you pay attention to the fact that from TF=60 min the relation between adjacent samples is almost zero (see your fig.), i.e. Markovian process, starting from that moment, is born for EURUSD into Wiener one. Isn't it interesting? People are working on the "days", but there is nothing to do here, starting from one hour!