a trading strategy based on Elliott Wave Theory - page 255
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
Haven't been much luck with "New Year" - while I was getting home after the holidays I slipped - double fracture of the ankle with dislocation - the cast was removed recently - so I have limited movement so far. I had to drag everything home from office when I began to get up, and the connection from home is only GPRS, and it is constantly adjusted - so it does not work at all.
So there is not much news yet. I tested the Expert Advisor in different variants and unfortunately since 2004 the general profitability is not so high as I would like it to be and the drawdown is 40%, which I think is unacceptable for the real account (although it was in the real account and in the tester this period is normal (the second half of 2005 and 2006 up to August and then near October passes perfectly), but the last drawdowns still leave some doubts). So I am still experimenting. The only problem with the Internet is that it works some and unstable.
Good luck.
All the girls, if there are any among us, I congratulate you and wish you success in the field of trading.
I have finished my model experiments with renko-kagi constructions based on Pastukhov's dissertation and I am posting the results of gross return comparison for 2006 and 2007. It can be argued that with almost identical returns the kagi constructions have almost twice as much return on the same historical data. This is due to the higher number of transactions per time unit. Therefore we will only consider the Cagi-builds below. The figure below illustrates the stability of the gross income for these constructions. Circles show income in pips for EURCHF and EURGBP in 2006, squares show income in 2007.
For illustration, I present the similar data on the autoregressive model profit:
The parameter predicted by the AR-model is plotted on the abscissa axis.
One can ascertain the superiority of AR-constructions for these symbols on this tick sample.
Unfortunately, I was not able to get a statistically reliable superiority of profitability over the spread, even 1 point, neither by the kagi-build method, nor by AR-builds for EURUSD.
Personally, I stop studying the material provided by Pastukhov as less promising compared to autoregressive models.
You've been quick.
I've only read it and ran the H volatility for a while. I understand that kagi builds are not very profitable. But there are also multi-step models, as Pastukhov says, i.e. paternas. The most beautiful thing is that if there is a measure introduced by Pastukhov, then neural networks are not needed at all, you can search for patterns and classify them using this very Pastukhov's measure, no re-training and other problems related to networks arise.
But I haven't got to that yet. Stuck on channels and Hurst. I think it's still promising.
My applause!
This is probably the most valuable observation I've ever come across. I have no doubt that it will make you happy and rich.
Try the signal
Bid<Low[1] = buy
Bid>High[1] = sell
I think it will give you a hint.
Good luck!
Reading the material, I reminded myself of a character from the cartoon "Investigative Behind Bugs". He stomped his feet and periodically said "I don't understand anything...". :о)
to Vladislav
I am glad that you are with us. Get well soon, there are a lot of interesting things to look forward to. :о)))
to Yurixx
Sergey, it seems to me that this is an incorrect formulation of the question.
It should be assumed that the 5th count is counted from the current one, while the current one is zero?
It is clear that they will not coincide. And this coincidence can't be for any iterative procedure that uses a finite piece of history of different lengths. ....
Perhaps, I have not expressed my thought very correctly. I meant in relation to a fixed current bar. For example, for potential energy of a channel in a sample of 500 will exactly sample 100 of the current fixed bar, but it won't be the same for the Hurst index. All in all, it's not very important. As I continue my research, and remembered, about the potential energy of the channel...
***
I decided to share the "technology" and "advanced" criterion of determining the reversal, which I also use in my model of strategic forecasting. It works well on both small and large samples (in relative sizes, of course). Keep in mind that I am presenting the material in a very conceptual way.
Let me remind what Vladislav shared earlier: the price is most likely to develop within the bounds of the channel having the minimal value of the potential energy. Correspondingly, this minimum may occur at any place of the sample of length N. So I wondered once if I could somehow estimate the dynamics of development of local extrema (in my case I need to pay attention not only to the lowest one) and make assumptions as to the possibility of a reversal based on the dynamics.
The number of samples N=100, the "dead zone" is 30, and the potential energy (PE) of the channels is as follows:
And this is how the calculation of PI one bar backwards with a window of the same size looks like:
Assembling the "dynamic model" is quite easy:
Another angle:
The arrow marks "now", i.e. taken as the current readout. The "WINDOW" axis, is the window, long N, through which we "look at the process". By estimating the PI level and its relative change, we can make assumptions about possible reversals. So, for example, if there is a steady downward trend of PI levels of local extrema and their relative shift closer to zero, there is likely to be a reversal whose scale relative to RMS of channels should also be estimated, i.e. a "bounce" from a boundary or departure from a channel when there is "leakage" of energy into neighboring channels.
Of course, it is not necessary to build such a model, but I need it for additional calculations (this is far from all that can be "pulled out" from such a model). We can estimate as well (calculation of EP for a sample of length N from the current bar and EP of length N counts back when the current bar was -N is shown)
This is the kind of reversal we are seeing:
Very interesting, but unfortunately nothing is clear. :-))
Mainly probably because it is not clear what you call potential energy. And since it's not clear what that quantity is and where it comes from, it's also not clear how to interpret what you see in the pictures. The traditional interpretation (the system moves along a phase trajectory, which minimizes the potential energy) does not seem to fit very well either, because it contradicts your own statement:
And from the classical point of view it is vice versa: the lower the level of potential energy, the closer the system is to the stable equilibrium position.
Some other points are not clear as well. For example, the first picture is preceded by the phrase:
What is "dead zone" and why does it say "potential energy (PE) of channels" if there is only one curve in the picture ? Or do all channels have the same potential energy ?
I warned from the beginning that the presentation is conceptual in nature. :о)) And mainly pursued two goals:
(1) to trace the dynamics of PE behaviour, well this is for those who calculate it somehow, and to share the extended PE criterion
(2) to talk about the "dynamic model" used. This simple method can also be used for other indicators, not necessarily for PE.
In general, of course, I should have written a bit more, because really not everything can be understood. I thought that a hundred and more pages contain what is necessary. But, come on, don't erase the post :o)))
The principle of channel search is the same as for Hearst's index, about which I wrote earlier. Nothing new here. The rightmost bar is the bar taken as the current count from the whole sample (remember, about 50,000 entries). The sample values are (H+L)/2. The "dead zone" of 30 samples is the first channel for which PI is calculated (it is the first point on the right of the red line symbolizing PI). This is the minimum at which it makes sense to calculate anything at all. Then, at +1 step I go to the history from the "dead zone". This (+1 bar in the history) iteration limits a new channel for which the second calculated point of the red curve on the right appears. And so on, until the end of the sample and until the end of the red curve line ("end", already on the left, we went from the N-zone), which symbolizes the EP.
Fixing the length of the window, we look what was one bar ago (as if transferring the reference point, in other words we can stand on the roadside and look at the car, or we can sit in the car and look at the road, yes, yep, you all know it better than me!!!). Correspondingly, each point on a PE surface corresponds to some channel [0; n] in the range [0;N]. Channel [0; n] is a "chunk" of the price, for which everything is calculated, that's all.
We've been thinking about PE for a long time, starting on page 4, and you remember it well. Everyone calculated it differently, someone "bent the beams", someone else had fun with it somehow. I've been looking for it for a long time, found it and am sharing its "look". On the whole, it works and really has a quadratic view. Stop! Here's more details. For each specific channel (sample price in an iteration) the PE (each point on the PE curve) is a quadratic form, i.e. each point on the red curve is calculated using the quadratic form. (The point on the PE curve corresponds to the price/channel segment) My PE works fine, and I don't even ask Vladislav about the similarity. It works the same for me in conjunction with SME :o)))
Honouring the earlier rules and remaining silent about the PE calculation itself. :o)
And from a classical point of view it's the opposite: the lower the level of potential energy, the closer the system is to a stable equilibrium position.
And what makes you think it contradicts the theory, given "it's not clear what you call potential energy". Just kidding! Actually there is no contradiction!!! The equilibrium point is slowly shifting closer to the origin(not always, sometimes the opposite process is observed, and why should one think that the PE must always stay in place?), just so that this equilibrium is not disturbed by the coming reversal. An analogue (I repeat, it is only an analogue that has no direct relation to the process under study) - a motorcyclist when making a steep U-turn, in order not to fall he practically leans to the ground. I.e. at first he drives straight ahead (equilibrium point takes any value from equilibrium space), then he starts to turn, changing his equilibrium point based on the prevailing conditions (speed, mass, friction, etc.). If it doesn't do this, it will just keep on going..... can't turn.
In other words, the system is rearranging itself before your eyes, changing its characteristics. You can't tell which way it will "wiggle", there are other indicators for that (this is me talking about channels).
Addendum...
In conclusion: the following correlation is true: a channel limited by a price sample [n:0] corresponds to one total value of potential energy of all constituents of movements in the channel. Here.
PS: changed the picture for clarity
As far as I now understand, the PE value at a given point changes as the current count moves more and more to the right ? After all, the channel that connects this point and the current datum is changing.