a trading strategy based on Elliott Wave Theory - page 191

New comment
 
<br / translate="no">Alien,
Neutron You said you have been doing neural networks. How do you think they are applicable to forex forecasting? And if there is any perspective on their using in forecasting? After my attempts (not without success) to reproduce Vladislav's system I have started dealing with neural networks.

Hello, Alien,
I do not deal with networks as a matter of principle, because I can show from general considerations their lower validity for financial instruments in comparison with autoregressive models. My personal, unfounded belief is that the use of this apparatus is preferable only in one case: if we understand ANYTHING at all about the pricing mechanism. But, sorry, that's like shooting at sparrows.
 
Sorry for the flood, but I couldn't resist. According to the TV show "Let them talk" (ORT) the aforementioned mathematician Perelman lives in absolute poverty and works as something like a longshoreman in a vegetable shop. Not only did he not have the money to go for the prize, he simply was not allowed to do so. In his place, some random bureaucrats or thugs came to the ceremony hoping to steal the money. When they (of course) refused, they declared Perelman insane and said he didn't want to get a million quid. Woah-woah-woah.

Unfortunately, Russia has been using American "technology" in the media for a long time now. That is why there is so much misinformation and dirt in them. And, to make matters worse, human rights technology and the rule of law leave a lot to be desired. That's why one can badmouth EVERYONE with impunity.

Perelman worked (and I believe taught) for several years in the USA. He published his work there on the web site of the university where he worked, as a preprint. He did not even bother to submit it for official publication as a scientific paper. This already speaks of his attitude to the officiousness.

Then he returned to Russia, despite offers from Americans to continue his career in the US. This, too, says something. The Russian Academy and institutes, where he could have continued to do science, were not interested in creating the conditions for him either. And he is not interested in doing anything but his research. So his poverty is his conscious choice. And it does not weigh on him, otherwise he would have no problem finding money to go to America for a million and solve this question once and for all.
Which is exactly what I wrote about.
 
Hi all!
I found a small program on the Internet http://www.matrixer.narod.ru (1.4 Mb) which allows you to build a spectrum for an arbitrary time series, and many other things. I think it will be interesting to play with. If my dear Rosh shows me how to attach a file with quotes archive, I can share some EURUSD Dy for the last 20 years.
For the sake of example I am showing the result of spectrum calculation for this pair. On the horizontal axis the frequency f is plotted.

To go to time t, you need to take the inverse of it: t=TimeFrame/f, where TimeFrame, in this case, is a day.
 
Go to forum.mql4.com/en , for example, to the "MQL4: Metaquotes forum picture" thread
And attach a file in an allowed format.


And in this branch, you will report the existence of such an uploaded file and give a link to download it from the forum.
 
Thank you.
Here's a link to the zip archive with the two files:
"MQL4: Picture for forum on metaquotes".
After unzipping, you need to run Matrixer on them and you can watch the spectrum of the series. The dUSD series has the following algorithm:
X[i]=Open[i-1]-Open[i]. This procedure is done to centre the series, which is necessary to calculate the spectrum correctly (triangle button).
 
2 Neutron
Could you please explain what number series centring is.
Thank you.
 
Centring is used to average the price series. For example, series A[i] (i=1;n) is replaced by series Bi=(A[i-1]+A[i]+A[i+1])/3 . An example can be found in Schwager's "The Complete Course of Technical Analysis".
 
Thank you, Rosh.
However, I don't think that's it. I asked the question because Neutron repeatedly referred to centering in his posts as an operation that precedes spectral analysis of a numerical series. And in his last post he wrote
The dUSD series is constructed using the following algorithm: X[i]=Open[i-1]-Open[i]. This procedure is done to centre the series required for correct calculation of the spectrum

As you can see, it is quite different from the variant you have given.
That's why I would like to understand what centering is. Better yet, I'd like to know the strict mathematical condition to which a numerical series must obey.
 
Centring is done to filter out random outliers in the price series.
 
I asked the question because Neutron repeatedly referred to centering in his posts as an operation that precedes spectral analysis of a numerical series. And in his last post he wrote
The dUSD series is constructed using the following algorithm: X[i]=Open[i-1]-Open[i]. This procedure is used to centre the series for correct calculation of the spectrum

...
That's why I would like to understand what centring is. Better yet, I'd like to know the strict mathematical condition to which a numerical series must obey.

As far as I can see, many procedures (e.g. Hearst ratio calculation) are only correct for samples with zero expected payoff. Exchanging price data to X[i]=Open[i-1]-Open[i] just gives a sample that barely meets this condition. In a sense it can be called centering, perhaps this is the conversion effect that is meant. But another question arises: Doesn't this transformation also produce a certain randomization of the original number series?
1...184185186187188189190191192193194195196197198...309
New comment