a trading strategy based on Elliott Wave Theory - page 171

New comment
 
Hi all, I haven't read the whole thread here, but I was just wondering if the EA managed to implement the system in question in the beginning? :))
 
Hi all, I haven't read the whole thread here, but I was just wondering if the EA managed to implement the system in question in the beginning? :))

I think many have implemented it, but not so many have had good results. For example, Sollandr has succeeded in something. The last post on this page reads : "Trading strategy based on Elliot's Wave Theory".

Maybe he posted the results later, but I do not remember where.
 
These results from the system discussed in the beginning are indeed the latest. Unfortunately, this system had to be abandoned due to the noise dependence of the Hearst index calculation, which is the basis for market entry decisions. In this post at the end of the page you can find the detailed explanation of problems with calculation of this parameter. In brief, at different brokers at the same time values of Hearst Ratio calculated with the same algorithm will be different, that is non-critical when the ratio is far from 0.5 and more than critical when the ratio is close to the area of 0.5. Hence, the performance of the algorithm based on the Hearst's index differs on the quotes of different brokers. I was not satisfied with it and completely abandoned the use of the Hearst index calculation.

I am now doing the system described in this post:
"trading strategy based on Elliot Wave Theory" solandr 04.10.06 10:11
From this post there are links to its description further
"trading strategy based on Elliot's Wave Theory" solandr 27.08.06 21:05
"trading strategy based on Elliot's Wave Theory" solandr 08.07.06 20:12
Judging by the results we got for the last 2 months both on demo and on real this system has higher noise stability as the opening of orders takes place at calculated levels and the system makes absolutely no difference in what way and at what moment the price appeared at these levels. When comparing orders at Alpari and InterBankFX, the difference in calculation of levels can be up to 15 pips (I did not calculate the average difference, but I think it is about 5 pips) It does not fundamentally affect the overall picture of the system.
Unfortunately, it is not possible to present the data of strategy run in an Expert Advisor, because strategy operation is in semi-automatic mode. Market entry is done through pending orders placed by the Expert Advisor. In the majority of cases I exit manually, though the Expert Advisor has all possibilities for closing positions early under certain conditions, not only using Stop Loss and Take Profit, i.e. it is a fully functional Expert Advisor in terms of position management. Of course, it is more the psychology of trading (my nerves are far from being stable seeing an increased probability of price reversal against a profitable position and waiting for a pullback to continue movement. But who knows what is better - nerves of steel or "bird in the hand"? ;o)). I prefer to make a profit, even if it is small, rather than wait for the objectives to be reached. Although, perhaps half of the time the objectives are achieved. Well, I think I need to improve my strategy in terms of additional confirmations for which a profitable position should not be closed. In general, we are not sitting still ;o).
 
I froze this topic at my place about two months ago. The motive is that the calculation time is too long with too many factors. I cannot correctly choose a variant from the first principles, the tester test will require an unacceptable amount of time. The best of the options I have found did not give much profit, at the same time I could sit in zero for two or three years. Perhaps I will return to this approach - if there are ideas how to facilitate it correctly and there will be additional clarification about the first principles.
 
What about your EA that worked well on the demo (or maybe even on the real one)? Or the strategy with a quadratic regression is still under development and its implementation in that EA is not finished yet? <br/ translate="no">
If I'm not mistaken, the quadratic channel has the same interpretation as the linear one or there are some pitfalls?


I have an Expert Advisor and I traded not only on demo accounts. The main problem with the linear channels appeared in August when the drawdown started to grow at a higher rate than expected. According to my research (almost till the end of September), this was caused by restrictions in the number of nesting levels. Which was a forced measure due to the size of computational costs. Removing the restrictions, however, leads not only to an increase in computational costs but also to noise-dependent signals. I am currently investigating a quadratic algorithm. Computational costs are less than for linear channel approximation due to the fact that many of the criteria are automatically satisfied and most of the tasks can be abandoned. I think the logic of use itself should not change (at least I use, or rather try to use, them too. So far, I've used them manually - I don't risk to leave them for untested EAs). I haven't finished "fiddling" with algorithm convergence and determining the optimal (in the sense of sample size and convergence speed) criteria. I hope it won't take long to restart the Expert Advisor using the selected criteria.

Regards, Vladislav.
Good luck and good trends.
 
<br/ translate="no"> These results from the system discussed in the beginning are indeed the latest. Unfortunately, this system had to be abandoned due to the noise dependence of the Hearst ratio calculation, which is the basis for market entry decisions.


solandr, do you remember our 30-page discussion about the Hearst Ratio? If memory serves me correctly, you don't calculate the Hearst index...
 
<br/ translate="no"> solandr, and remember our conversation about the Hearst exponent in 30 pages. If my memory serves me right, you don't calculate the Hearst figure...

I think the classic calculation of the Hearst exponent would give the same noise-dependent result.
 
[

solandr, а помните нашу беседу о показателе Херста на 30 страницах. Если мне не изменяет память, Вы как раз и не вычисляете показатель Херста…

I think a classic Hearst index calculation would give the same noise-dependent result.


I don't calculate the Hurst index using Vladislav's algorithm, but my calculations show that using the classical definition gives just as good results. But here also the experiment will not be "pure", in the sense that I will use other competing criteria to select stable channels. True, so far all the calculations are in MathCAD.
 
IMHO don't mess around, no clear conditions no clear implementation, Elliot's wave theory is more of a philosophy
 
IMHO don't mess with heads, no clear conditions - no clear implementation, Elliot's wave theory is more of a philosophy <br / translate="no">

Thank you very much for the recommendation! ;o)))
1...164165166167168169170171172173174175176177178...309
New comment
Reason: