a trading strategy based on Elliott Wave Theory - page 133
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
In the strategy, as one of the supporting factors for the reversal, it is probably still appropriate.
I agree. If I check it out, I'll let you know the results.
As part of the strategy, it is probably still appropriate as one of the supporting factors for the reversal.
So I'm talking about the criterion - if the criterion confirms the need to consider the parabola (its relevance on the chart) - then we can check the vertex passage. If it doesn't - we ignore it altogether. As a criterion, I think Fisher will do.
I think Murray levels are more reliable than Fischer;o))).
I think Murray levels will still be more reliable than Fisher ;o))).
Don't be so sure :) Besides, Vladislav has made statistics for Murray levels, while for other methods (Pivot was mentioned) statistics must be recalculated.
In other words - everything that is not forbidden is allowed. The only thing forbidden is fitting.
Initial deposit size was chosen as 100USD to bypass server limitations on maximum lot size when working with large balance at the end of testing. The risk limited for 1 trade is 25%. At this risk tolerance, the maximum relative deposit drawdown was almost 40%. I also ran the tester at other values of the acceptable risk and got the following information on a test sample of this modification of the trading algorithm. At a risk of less than 25%, the graph of the deposit growth is qualitatively the same as the one shown for 25%, but the resulting profit is less. At a risk higher than 25% (30%, 40%) there are additional 1-2 drawdowns of the deposit (usually in areas of reversal of medium-term trends, when there are strongly opposing views in the market, which have an equal chance of existence), but the balance grows faster (have time to withdraw before the drawdown happens :o)! So far I've decided to stop at 25% risk. In my strategy this is the only adjustable parameter. Everything else is determined only by the entry-exit algorithm. That is why I have already tried a lot of variants of its implementation based mainly on graphical methods.
The results of the previous modification of the Expert Advisor in real trading were vague because of a small technical mistake made due to my inattention and there is no reason for me to publish them here. The mistake was in calculation of the lot size for deposits less than 1000USD. I've always tested the strategy for an initial deposit of 1000USD and somehow managed to forget that I only have 600USD in my account. As a result, because there is a close relationship between entry point, lot size and allowable risk in the EA, I simply missed about 30-40% of entry points. It took me about a month (!) to understand what the problem was - why the EA did not work with some entry points according to the trading algorithm :o(. But better late than never. As a result, last week I ran the Expert Advisor with corrected error in lot calculation (you can see the results on the history). I have specially started tests with very small initial deposit to avoid problems with money management with small initial balance.
At about the same time I had an idea to build channels based on convergence points or equality of the sum of gradients to zero. Briefly, the essence of the idea is as follows. For every point in the history we build channels of linear or parabolic regressions considering this selected point of history as the beginning of the channel and further bars (the bars that appeared after this point) as the end of the sample, i.e. those which are closer to the current time. For each of the samples obtained in this way we build channels, find differences between the regression line and, for example, the opening price of the bar of the end of the sample, i.e. find a gradient for the channel built in the sample. Then we sum up the obtained gradients. These sums of gradients for each point of the history are shown in the charts below for illustrative purposes.
As you can see from the figures, each point of the history has a different value of the sum of gradients of channels built on it. Further we assume that the points of history where the sum of gradients changes its sign (is close to zero) can have some prognostic properties and the regression channels built on their basis can be useful for trading, presumably in the short term 1-3 days. To demonstrate the results of this technique I have laid out a brief variant of screenshots by days here https://c.mql5.com/mql4/forum/2006/08/EURUSD_800_600_small.zip 500Kb
And also full version of channel development for the period from 01.01.2005 till 26.08.2006 here http://et385.narod.ru/forum_metaquotes/EURUSD_800_600.zip 19Мb
I apologize in advance for slowness of downloading files from folks.ru. If the MQL4.COM administration allows me, I can upload this big file to MQL4.COM for normal downloading.
Based on a month-long observation of the channels built using this methodology, I can report that it is likely to be useful when playing manually as a pretty good confirmatory indicator for intraday play. That is, we build daily channels and then decide during the day how likely the movement may be in one direction or another. When playing, we use both channels borders and central regression lines.
Unfortunately, I was not able to formalize in any way the algorithm for creating an EA viable only for this method. But maybe it is only for now. If anyone manages to download the full version from people.ru, you can view a slideshow (animation) in ACDSee which will suggest the existence of moments of time when the forecast using this method was made before the price moved. In any case, the method as it stands now needs serious refinement.
PS: Also on the charts (on the above links, one of which is also given here) there are the averaged points of the borders and the central line of the "averaged regressions", which shows a higher order trend, which with some probability, can be justified exactly by the economic factors occurring in the global economy. And those oscillations that the price makes around this "fundamental trend" for example under the influence of some news are speculative fluctuations of the market, due to which the traders can earn money.
PS: At mql4.com forum "MQL4: Picture for metaquotes forum" I posted the complete file which is located at http://et385.narod.ru/forum_metaquotes/EURUSD_800_600.zip
Multivolume RAR archive. There are 20 parts in total.
After downloading all parts, change zip extension to rar and unzip to WinRAR3.50. (Just forum does not allow to upload files with rar extension.)
I had to cut it into 1Mb chunks to fit the forum limitation. Apologies in advance to the forum administration for hosting so voluminous. Just thought, that in any case, firstly, it is not lost somewhere with time, and secondly, it is unlikely to raise too much traffic from server, since it is interesting only to a very small circle of traders, who understand that the happiness is not in MagicNumbers and not in the windows opening / closing positions, but in something completely different ;o))). Well, if it will be a strain on the server, then you can remove these files at your own discretion.
solandr, did you run your EA on pre-2004 data? ? The thing is that on my data since 2001 the images look like this
I.e. since 2004 it's favourable period, and before that time it was useless in fact. It is clear that we have different implementations of the method (and I haven't used MM yet - I don't think the quality of inputs is good enough).
The broker's server has M30 data only from October 2004. I haven't looked for additional quotes specifically yet. Although, it would be probably worth checking the Expert Advisor on other historical data as well. But where could I get those quotes? The broker refused to provide me with M30 history, despite the fact that I have a real account with him. Basically, from his side - it's probably just a lack of respect for the client, even though he may have switched to MT only in 2004 and could not accumulate quotes M30. But he has quotations of Н1, Н4, etc. for a much earlier period of time, doesn't he? Didn't he get them somewhere?
Well, having a history centre for MT4, which the developers are now working on, is a solution to this painful issue that faces ALL creators of MTS. And probably not only M1 should be available on this centre, but all other periods as well. The good thing is that the other periods are much more lightweight compared to M1.
Perhaps you could purposefully post those quotes at MQL4.COM? Probably in 2-3 parts you could post them on the forum.
As for different periods, the standard period_converter script allows for quick preparation of all necessary minutes. So having the minutes for a long period completely solves the problem.
I think the problem may also be that over the last 2-3 years the very amount of EUR currency to play in the market has increased, because people pay attention to it, trying to find a reserve (safe haven) currency, the role of which has long been played by the dollar. As a result, the turnover and the number of market participants increase. And accordingly, the "statistical" character of EURUSD increases. (It should be remembered that the EUR currency itself appeared recently and had a period of uncertainty in the beginning (its appreciable fall after its introduction). For example, my EA on USDCHF and GBPUSD shows more modest results, than on EURUSD. You should agree that on smaller currencies there is VERY much more possibilities to move the market as you wish than on EURUSD. There are very few market participants who can do it on EURUSD, though on other currencies it will be much easier. That is why theoretically EURUSD should be the most profitable one for applying methods of mathematical statistics. But this is just my opinion.