a trading strategy based on Elliott Wave Theory - page 181
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
Basically shared with you, in confirmation of my post on page 86: "grasn 13.11.06 19:07", remind me what I responded to:
Unfortunately this system had to be abandoned due to the noise-dependence of the Hearst index calculation, which is the basis for the market entry decision.
I've spent many months researching, and I can assure you that Hurst works fine. The initially "twitchy" indicator after calculation is normal, it inherits fractality if I may say so "at the macro level" from the original signal
And one should not determine reliable channels based on competitive criteria, this is a major mistake One should go from refining the channels obtained, which are not very many. Always, one of the channels found (and it doesn't find that many, from 7-20 from a sample of 700-1500 samples) hides a reliable channel.
Rosh
If you mean the example described in the post "grasn 02.12.06 01:38", with these channels:
----------------------------------------------------------------------------------------------
Extreme...........Counting..........Hurst......................Channel length
-----------------------------------------------------------------------------------------------
[1]...........................287........................0.909..........................413
[2]...........................379........................0.792..........................321
then it looks no different from other similar ones with H>0.6. That is, it looks nothing remarkable, it is just one of the spaces where events can develop "folding into channel", repeating folding
You have to understand that this is certainly not a panacea, not a Grail cup. Yuri is right that we do not obtain true data but only get closer to them and as a consequence a channel with H=0.792 is more reliable than one with H=0.909. It is necessary to investigate calculated channels as a whole, for this purpose I also gave channel length for reference and the pictures show borders of 1*SCO. I gave the readings for better orientation on the charts as they are not very similar to MT and may not be as readable.
How to recognise (rather than how to find) this is the most interesting part. I won't reveal how I do it (I've practically learned already). I will only say that I have shown - this is a static model, and there are also dynamic and refining parameters.
PS: after rereading, I realized that probably, the material was not given very concisely and clearly, I thought that this topic is known to all
I'll share a bit about how I set the task and what I was looking for. The figure shows a general case of a channel reversal. I can see a lot of such elements on price charts at different scales as well as its complete opposite. The most important zone in this chart is A. The prices in this zone belong to two channels at once. The objective was to detect an incipient channel in such zones.
The same reversal can be seen on the chart in my example:
It is interesting that Hirst found such a structure and gave it a high score of 0.792. No wonder, the new channel was already "sitting" in the old one. No trick, I confess I intentionally moved the current bar a bit to the right, sorry, it was to enhance the storyline. :о)) Up to 300 bars, Hearst would not suspect anything (tested). Given that it's an H1 (hour) period then from 300 bar to 700, er how much would that be ... out of 700-300, exactly! 400 hours!!! Wouldn't that be enough time for you? :о)
Considering the neighbour's 0.909 and the accuracy of the calculation,(and given its reversal nature: all showing up and one showing down) you can at least give this extremum the attention it deserves. At least keep such a development in mind and track it. Don't forget about my rule:
(1) Each of the channels found already has sufficient stability (in general, the subsequent data holds within 1-1.5 SSR, and even more so at 2*SCO, and keeps its structure) for H values close to 1.0 (or a bit higher). For values close to 0.0 an early reversal of the established structure is confirmed.
Once you enter any channel (you can increase the limits to 2*SCO to be cautious), you will have time to get off and "change trains", unless of course you do something outright stupid.
There is nothing special about digital filtering (I briefly described my approach, based on DSP, in a neighboring thread). You can filter everything at once, you can filter what is called real-time. In this case I filtered "all at once" (in brief):
1. The whole Hearst index signal is taken
2. Perform Discrete Cosine Transform (DCT) for the whole signal
3. The resulting DCT removes the noise component
4. Perform inverse discrete transform
This filter has one disadvantage - it has a boundary effect that slightly distorts the signal at the very edges. For the examples I did not use more sophisticated approaches to digital filtering, which I have, for the simple reason that in my model I will not filter it right after calculation. The filtering will come later.
It's not about filtering the indicator at all, it's about which Hirst calculation model to use. This is the main thing.
Since everything is so clear in your research plan, could you also perform a comparative analysis of the "classical" calculation, which you use, with the Hurst index calculation according to Vladislav's scheme? You can use the same data as in the posts above. I think it shouldn't be too hard for you? And this comparison of different Hearst ratios may be a tangible contribution to the work on the applicability of the Hearst index to the capital markets. Maybe it will appear that they can complement each other, or for example filter out false channels? Maybe in the future you will want to write an article on Hearst indicators at the same www.mql4.com for example?
Дело совершенно не в фильтрации показателя, а в том, какую модель расчета Херста использовать. Это главное.
Since everything is so clear in your research plan, could you do a comparative analysis of the "classical" calculation, which you use, with the Hurst index calculation according to Vladislav's scheme? You can use the same data as in the posts above. I think it shouldn't be too hard for you? And this comparison of different Hearst ratios may be a tangible contribution to the work on the applicability of the Hearst index to the capital markets. Maybe it will appear that they can complement each other, or for example filter out false channels? Maybe in the future you will want to write an article on Hearst indicators at the same www.mql4.com for example?
It certainly sounds tempting. Let me remind you that my posts are expressing my own opinion and I am not agitating anyone for anything. I'm just sharing some of the research I completed about three months ago.
This is the comparison I was making. If I'm not mistaken, Vladislav calculates Hurst index as follows (at least based on forum materials):
H=log(R/S)/log(0/5*N)
Where R=High[]-Low[]
And for S calculation Open[]
The model is rather crude (a priori not suitable for price series), while the data selection is not at the very core of RS analysis (mind you, this is my own opinion and understanding). Ended up settling on the "classic" Hearst.
I am not pursuing such large-scale goals as making a contribution on the direction of applicability of the Hearst indicator in capital markets. For myself I have worked it out.
You can do similar calculations on your own, and write your own article. :о)
Alex, how do you think things will develop further?
I confess that I thought that till the New Year the Euro will roll back once more, but not much, and the break-up of 1.30 will occur right after the New Year. It is a bit of a surprise for me. What do you think?
By the way, November is over. How is your demo account doing?
Alex
But seriously, it's 17:45, the current quote of EUR/USD is 1.2714,
waiting for 18:00, I want to sell EUR at 1.2785, in 2 months I will close at 1.2400.
This is a significant statement ! I am not suggesting a bet, but I would like to put mine next to your statement
.
EURUSD might be down one more time, but hardly below 1.2500. And it is unlikely at all.
In 2 months, i.e. by the New Year, it will most likely be around its ceiling of 1.3000.
Let's see what gives more to whom, you EWT, or me my analysis tools.
Not as a competition. It is just that you have stated so unambiguously in favor of ЕWТ and
have convinced us that you are an expert in it and that it is with its help that you have achieved those
fantastic results, about which you wrote at the beginning of the branch. That is why I would like
to compare its capabilities in the hands of an expert with my, quite amateurish, predictions.
Good luck.
Yuri, I assume that's what you mean? Right, kind of a surprise in a way. I'm not trading now, but, as I wrote, I'm doing research (sometimes it's very useful, remembering the words of one of the greats "you don't have to trade all the time"). But just in case I checked, my Hearst is showing very unstable "long" channels, and portends a fall soon. Admittedly, my "energy" method based on Hearst's indicator is not yet fully worked out.
PS: where did Alexa get such success? Working without stops he should have lost almost everything on that trade. Although, on the other hand, if he made another two or three million, that's not terrible at all.... mastery, though!
Actually no. The time for that bet isn't over yet, and summing it up ahead of time isn't serious.
Not to mention the fact that that bet has more of a humorous connotation than a principled one. You can't judge anything on one occasion. Not about EWT, not about how Alex uses it, not about my abilities. I'm against it.
It's just that Alex, at Jhonny's request, wrote
And now, after this snark, and I'm wondering how Alex has worked out in this situation.
As for the prediction, I can see that the eu still has upside potential. Probably to 1.36 or less.
So, if there is a correction now, it will not be too deep. 150-300 points