a trading strategy based on Elliott Wave Theory - page 251
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 doesn't matter who uses what, it's the result that counts.
I doubt that Lomonosov invented a table trying to imitate someone, or even copy someone... :о)
networks, mappings, diffusers, wavelets, fractals... there is truth in everything!
and I'm curious about something else today: http://www.berdyansk.net/club/mist/nom_1/myst_m.htm
:)
you could go by Demark, for example...
And Pastukhov seems to have disappointed even the few interested. And for nothing!
In the end the discussion stopped at the most interesting place: how to make a real working strategy out of purely mathematical
results. Well, who needs it?
Solandr:
You can be sure that EVERYONE is interested! And Pastukhov and your research.
Yes, I'm the one who "cast the fishing line". The discussion of Pastukhov's thesis began smoothly and ended abruptly. I read it with great interest, despite my attitude to the subject. So I'm curious to find out the outcome :o)))
Solandr:
And why don't you use as red dashed lines not RMS, but borders of confidential intervals constructed on the basis of criterion of 3 sigmas, or calculated on Student's for example for a confidential interval of 99 %? Or do you have any special purposes for the chosen boundary construction? Just out of curiosity.
The choice of channel bounds, a very important parameter to pay more attention to than I have done so far. There is no particular purpose in selecting as channel boundaries the RMS at the moment, other than one - somewhere to start. The requirements to the borders are rather vague, they should not be too wide, otherwise the whole sense of the forecast will be lost, and they should not be too narrow, and in addition they should be as close to the High and Low levels as possible. By eye, I could be wrong, but the 3 sigma criterion will increase these boundaries. But I will definitely see what happens. Obviously, accurate prediction of (H+L)/2 will greatly improve the overall accuracy of the forecast.
I have very modest requirements for the forecast, even if the interval [High, Low] just "touches" the line, I consider that the forecast fulfilled (of course, this condition for each bar within the forecast interval).
I am more concerned now with the criterion of choosing the number of samples. So far, this method has a great disadvantage - there is no validity of the forecast except for the first criterion found. I will have to ask for help of autocorrelation and its convergence conditions which I wrote about long ago.
And here we go again with fractals and their skylines. Man, it's the same as for large samples. There is always a countdown separating some sort of 'structure' that is destined to repeat/extend/scale..... You could, for example, compare 14 and 71, though, what did I expect otherwise :o) Or maybe I'm fantasising?
OK, and "fractalisation" let's get on with it... Where to start? Oh! I'll try to single out the most "minimal fractal" taking into account that nobody knows the exact definition and generally what a fractal is. T-o-o-o-o, what does it all look like, worms of course!!! Square worms, I'll start with them. :о))) (that was a joke, no need to call an ambulance)
Sergey, I do not fully understand what is in brackets, but I can give you some advice.
Forecast High and Low separately. These two values behave differently and this is not by accident.
Combining them into one curve and defining a symmetrical corridor in this case will artificially
roughening the forecast. Why do you need it?
I should add. Even if you simplify your P(t) function, but separately predict
High and Low, you will get more tolerable results than you have now.
Hi Yuri. I meant that the condition of getting the [High; Low] segment into the predicted channel (in any variant, even if it just touches the channel a bit) is fulfilled for every predicted bar.
Predict High and Low separately. These two values behave differently and this is not a coincidence.
Combining them into one curve and defining a symmetrical corridor in this case is artificial.
coarsening the forecast. Why do you need it?
I should add. Even if you considerably simplify your P(t) function, but separately forecast High and Low, you will get more tolerable results than now.
Tried it, the calculations are not very reassuring. Oddly enough, choosing (H+L)/2 gives better results and in my mind - this is obvious. Separate High and Low are completely random values, while the range that price takes on one bar most likely makes more sense, at least I think so.
Yuri, what do you call more tolerable results? Do you mean a more accurate forecast? If possible, in pictures or formulas. :о))
I had my first baby! - It's a girl. Not much time for active bodybuilding on the forum yet.
Seriously, Pastuhov's algorithm shows profitability just above brokerage commissions and below the profitability of autoregressive strategy I have on my real account. I am currently trying to understand the existing potential of H-builds and outline ways to increase strategy's profitability.
Seriously speaking Pastuhov's algorithm shows real-time profitability just above brokerage commissions and below the profitability of my autoregressive strategy. I am currently trying to understand the existing potential of H-builds and outline ways to increase strategy's profitability.
CONGRATULATIONS!!!!! HOORAY!!!! :о))))))))
When I say forecast quality I mean the following:
The bars formed after the current countdown (vertical line) in my case fully satisfy the quality of my forecast. Everything else about finding a local extremum in a turning square will be done with additional logic.
Baaah Sergei ! What a news ! Congratulations !
It turns out that your body movements are effective not only on the forum. :-)))
Special congratulations to the mother and best wishes to the daughter !