a trading strategy based on Elliott Wave Theory - page 96

 
As for the EA, I have a principled approach. I'm not writing it yet, but I will do it only after I've finally got the right channel selected. In my opinion, it is not quite logical to judge the correctness of the channel selection based on the results of the Expert Advisor. It may be affected by, let's say, not entirely correct opening/closing strategy. So, I evaluate the selected channel visually by deviations and use my signal correlation function (it should not be lost :o), although it is not lost anyway, I still use it somewhere).
 
Maybe you have found what you need, but so far I have seen some discrepancies with Vladislav's hints. <br / translate="no"> Your statement:
You have obtained channels with confidence intervals. You have extended them into the future. When a lot of lines intersect beyond zero, what do you get? So this is the space where you need to minimise the quadratic function.

contradicts Vladislav's statement

So building projections comes down first to sampling, then to linear algebra.


He first selects samples and then extends them into the future (at least that's what he said) First we minimize and then build channels.


I do not see a contradiction. Vladislav does not seem to build trajectories of price movement (he does not extend anything). He has reminded me of that several times. I don't build it in the same way.

And I sample from the beginning and then use linear algebra.
 
And I start by sampling and then use linear algebra.


So I didn't get your idea right, let's think about it further.
 
И я с начала отбираю выборки и потом использую линейную алгебру.


So I didn't get your idea right, so we'll keep looking at it.


In the sense that it is necessary to select the samples, but there is no need to build anything evenly on them.
 
By the way, I've sorted out my Hearst index calculation. (I brought it up on page 30). It's now completely correct, now giving 0.62 for the whole row. It also turned out to be simple.
 
Yes, I forgot to add, a very important thing in my opinion. And maybe someone will find it useful, I don't know. Although maybe everyone does.

My script, in fact, has no limit on prediction length. I.e. there is a limit on minimum forecast, which is only needed to start calculation from [current bar - minimum forecast].

When I move to the left along the bars (go to the history), the algorithm keeps track of the maximum length of its existence for each channel (moves to the right without a limit) until the "channel destroyed" check condition is triggered. Thus, I measure the full length of a channel's existence from its "discovery" to its "termination". I get quite interesting statistics.

Accordingly, I measure all possible parameters for the channel.

PS: I'll have to put some OLAP to take a bird's-eye view of it, so to speak.
 
It's been a while since there have been any pictures. Current situation on the EUR and JPY



 
I love picture books too :)
 
Good afternoon everyone!
Saw this thread and became very interested. Probably hard to jump in a train that is approaching the terminal station at high speed :-) but there is no other way..... can someone summarize the information received during the work done? If not difficult and not sorry Maybe I can also mite something mite contribute. Thanks.....
 
Good afternoon everyone! <br / translate="no"> Saw this thread and became very interested. Probably hard to jump into a train that is approaching the final station at high speed :-) but there is no other way..... can someone summarize the information received during the work done? If not difficult and not sorry Maybe I can also mite something mite contribute. Thanks.....


Read page 16 of the thread.