a trading strategy based on Elliott Wave Theory - page 29

New comment
 
Privet,

Kakto podumal, a mozet liudi, katoryje xorosho znajut EWA skazet sdies' naskolko pravilny rulsy:

http://www.geocities.com/WallStreet/Exchange/9807/Charts/SP500-Articles/EWRules.htm:)
 
Dear Vladislav!

Sorry for bringing up the topic of iStdDev again.
You wrote:
If you approximate closing prices by a muving, then it should be the difference between Klose and значением мувинга на каждом баре.And you can use the standard algorithm - iStdDev( ).


But mql4.com has the source code of StdDev indicator and it doesn't calculate the sum of squares of differences you mentioned. It uses another function to approximate prices, the muwings are only used to shift it by one of its coordinates for each set of bars. That is, one offset for the entire sample, for which one indicator value is calculated.

Could you please advise whether you have evaluated the prospects of such an approximation for trading? Is it worth recommending to beginners studying statistics for trading?

Thanks in advance.

By the way, information for developers:
If you replace
ExtStdDevBuffer[i]=MathSqrt(dAmount/ExtStdDevPeriod)
in that indicator 
;


on

ExtStdDevBuffer[i]=MathSqrt(dAmount/ExtStdDevPeriod) - iStdDev(NULL,0,ExtStdDevPeriod, ExtStdDevMAMethod, ExtStdDevShift, ExtStdDevAppliedPrice, i-ExtStdDevShift );


then it turns into an indicator of deviation of values of the original indicator from the built-in one.
No significant deviation is observed only when ExtStdDevMAMethod=0 (MODE_SMA) and ExtStdDevShift=0 .
With any other combination of these parameters the graph differs significantly from the horizontal one at the zero level. Build 193 dated May 4 and May 18.
Please advise what causes such a deviation?

Now I have an opportunity for the forum hosts to express their opinion :-)

 
No significant deviation is observed only when ExtStdDevMAMethod=0 (MODE_SMA) and ExtStdDevShift=0 . <br / translate="no"> With any other combination of these parameters the graph is significantly different from the horizontal one at zero. Bild 193 from May 4 and from May 18.
Can you please tell me what causes this deviation?

Unfortunately, there is an error in the description of the iStdDev function. Shift and method are mixed up in the parameter list. We detected it only yesterday.

The correct description is "MQL4: iStdDev".
 
Dear Slawa!

After rearranging these parameters we get the following result:
ExtStdDevBuffer[i]=MathSqrt(dAmount/ExtStdDevPeriod) - iStdDev(NULL,0,ExtStdDevPeriod, ExtStdDevShift, ExtStdDevMAMethod, ExtStdDevAppliedPrice, i-ExtStdDevShift );



Now the graph is significantly different from the horizontal one at the zero level only when ExtStdDevShift is different from zero.
Did I apply ExtStdDevShift parameter correctly in iStdDev call?

What is interesting is that this construct also draws an uneven chart:

ExtStdDevBuffer[i]= iStdDev(NULL,0,ExtStdDevPeriod, 0, ExtStdDevMAMethod, ExtStdDevAppliedPrice, i ) - iStdDev(NULL,0,ExtStdDevPeriod, ExtStdDevShift, ExtStdDevMAMethod, ExtStdDevAppliedPrice, i-ExtStdDevShift );
 
Dear Vladislav! <br / translate="no">.
Sorry for bringing up the subject of built-in iStdDev again.
You wrote:
If you approximate closing prices by a muvin, then it should be the difference between Clause and the value of the muvin on each bar.And you can use the standard algorithm - iStdDev( ).


But mql4.com has the source code of StdDev indicator and it doesn't calculate the sum of squares of the differences you mentioned. It uses a different function to approximate prices and only uses a muwing to shift it by one of the coordinates for each set of bars.


In fact (Xi-X)^2 is equal to X^2-X^2
where Xi is the value on the i-th bar
X - average on these bars (for the period N)
X^2 - average of squares on these bars (for the period N)
 
As for the built-in indicator, I didn't evaluate it - I just looked at how the algorithm was implemented. I prefer to work with personally tested programs. As for the standard deviation - I have it calculated when calculating the mnc for creating a regression channel - therefore there is no point in recalculating it. I just save it as I calculate it for all channels. Then I select for those channels whose samples meet the criteria. The purpose is the same - to speed up calculations.

Good luck and good trends.
 
a request to you. Could you show me a picture of what it all looks like. <br/ translate="no">What does it look like for you?


With me it looks like this :



What you can see here is the frank markup. The confidence intervals are marked up from 60%, to 99%. Degree of nesting is 3 - smaller channels are not identified. We can clearly see the reversal zone. It coincided with the medium term trend (it is the widest channel pointing downwards) and with two nested ones I will call them short term and ultra short term - my classification. The regression lines are dotted as the Hearst's coefficients for the respective channels are <0.5. Originally, the indicator marking the price field was developed for hand trading - an Expert Advisor does not care how it is drawn there :).
So that you don't think that it's a tweak - there's an open order by the EA. I can see the time and price of entry. I am still struggling with the EA - similar entries should have been on EUR and GBP, although yesterday I have slightly moved the EA (and closed it too late, I might have missed the entry points). I see why I have chosen such a level of withdrawal of profit. At breakdown of the uptrend channel of the shortest trend is possible, but the strong pullback is also possible. Ideally, the Expert Advisor should recalculate everything correctly by itself. We will see.
What else I can recommend for real trading
- do not make decisions to enter the market if you are within the 60% confidence interval.
- Do not go short if below the lower 60% confidence interval, similarly for long positions,
but it seems clear.
However solandr has already stated it.

Good luck and good trends.
 
Thank you Vladislav!

I, too, prefer tested algorithms, or better yet, self-developed ones.

Dear Rosh!

Is the formula given by you (on the picture from the help, am I right?) equal to the root of the sum (1/N)*(Xi-MAi)^2 ?
Where Xi is the value (for example, Close) on the i-th bar,
MAi is the muving value on the i-th bar (in case of SMA it is the arithmetic mean of Xi, Xi+1, ... Xi+N-1).
 
Not equivalent, the root can only correspond to the root, I meant what's under the root.

Here http://www.alpari-idc.ru/ru/experts/article s/21.html I have made examples of the use (with checking) of technical indicators on arrays, the Excel files are attached.
 
Formula


[img]http://forum.mql4.com/c/forum/2006/05/stddev%281%29.jpg[/img]
1...222324252627282930313233343536...309
New comment