a trading strategy based on Elliott Wave Theory - page 15

New comment
 
OK, in principle, I can also write the formula by which I calculate
P=(O+H+L+C)/4;
delta=H-L;
Start_SELLLIMIT=P+kstart*delta;
Start_BUYLIMIT=P-kstart*delta;
respectively exactly the same for take profit and stop points. The same formulas, only instead of kstart you should use ktp for profit and kst for stop loss which are picked experimentally in the Strategy Tester for each timeframe separately, because the market structure differs a lot in different timeframes. Even a shift of the reference point of a candle can drastically change the given ratios.
This means, as I mentioned before, these formulas have exactly the same meaning as the formula of continuation S=V*t. According to the formulas given above, we get the midpoint of movement for the next candle P, while delta symbolizes our possible speed (or the possible distance which can be travelled for the next exactly the same period of time). Coefficients are selected in the tester for a certain brokerage company and selected timeframe. In fact, everything is very simple without complicated algorithms. The most common linear formula adjusted to historical data in the Strategy Tester, can also give a positive result. In principle PivotPoints also use exactly the same sense as the above formulas. My formulas are simply more simplified and logically understandable than various complex variants of PivotPoints.

Although I have an assumption that whatever linear formula you use to calculate start, stop and profit points - you will ALWAYS have the ability to adjust the strategy to the historical data on the tester. Try to implement your suggestion to calculate points based on the continuation formula and I think the strategy should show about the same positive result.

Basically, I even have this idea. We take different variants of calculation of points on different linear formulas, adjust them on historical data, and then we show the results of tests and say that we have made a strategy, which was not present in nature before!!!!:o) For example, we have invented SuperPuperPoints :o)))))) Accordingly, on this basis, we could even write a clever book of about 400 pages. Basically, people who are more quick-witted have always done so and will continue to do so in the future! Well, simpler people will buy and read these books.

By the way, what could you improve in this strategy? Share your suggestions.
 
The division is based on the readings of the standard indicator included in MT4, called Standard Deviation. <br/ translate="no">


Sorry for the late comments.
Here, if you don't mind, more details, at least in terms of methodology.
My point is: the standard deviation is considered in relation to the forecast. A mouving is taken as a predicted value in the standard indicator. Accordingly, the indicator of the standard deviation included into the standard delivery shows by how much the current price has deviated from the forecasted one by the moving average of a given order. In the case of trends, the prices may follow the moving average or the deviation, i.e. readings of this indicator may be anything and not influence the trend definition, like in flat markets (which, as I understand, is one of your key definitions of the market phases).
It's just interesting how you see the possibility of dividing the market phases by the deviation from the predicted price.

Good luck and happy trends.
 
Деление происходит на основе показаний стандартного индикатора, входящего в состав МТ4, под название Standard Deviation.


Just wondering how you see the possibility of separating market phases by deviation from the predicted price.


I don't think I can answer you any more than I already have. I don't have any particular methodology on the subject. I just thought that deviation shows market activity and I took it simply on the basis of experimental data. That is, if you do the same with the strategy, but without the use of deviation, the success will be either unlikely or insignificant.
 
Understood. IMHO - in the general case this is incorrect. I certainly use this parameter necessarily and it is one of the possibilities to get noise-independent estimates (let's call them that). This parameter is needed in order to estimate where you are in the confidence interval. Although, of course, the interval itself will depend on the type of distribution within (there are options to get around this - I've already written). In principle, to your strategy in terms of methodology, Bollinger lines are logically suitable for determining the values of the confidence intervals - they are built on the same muwings. The direction of the trend = direction of the moving average. However, this estimation will have a certain time lag. By using confidence intervals this lag can be mitigated.

Good luck and happy trends.
 
Indeed this strategy is somewhat similar to Bollinger. Of course, I don't pretend to be original at all, as I use what everyone else has on the surface. It would of course be interesting to see your way of calculating probability distribution. Now I do practically the same by simple multiple running of the system in the tester. That is to say, I determine those very maxima on the probability distribution function to set orders in terms of their success. If you can give a specific recommendation in formulas on how this can be calculated in real time while trading, I would be very glad.
 
To be honest, I can't get into the idea itself.

In my mind, it all sounds like limit-based noise fishing to me.
However, that's about how it's supposed to work. I don't know. I'll have to think about it some more.

Generally speaking, noise catching is a reasonable idea in my opinion.
Here is a primitive variant of its implementation. "MQL4: Tester's Error".
On 6 months' run it shows approximately 9000p of profit (spread 2, all ticks).
Ok. 1000 trades per month, i.e. approx. 50 per day, approx. 2 trades per hour. Quite a tolerable load for a dealer:)
 
Indeed this strategy is somewhat similar to Bollinger. Of course, I don't claim to be original at all, as I am using what everyone else has on the surface. It would of course be interesting to see your way of calculating probability distribution.


I use solely the integral approach where it is sufficient to find a convergent distribution - the type of distribution itself does not matter in this case.

Good luck and good luck with trends.
 
Generally speaking, noise extraction is a valid idea in my opinion. <br / translate="no"> Here is a primitive implementation. "MQL4: Tester Error".
On a 6 month run it gives approx 9000p profit (spread 2, all ticks).
Ok. 1000 trades per month, i.e. approx. 50 per day, approx. 2 trades per hour. Quite a tolerable workload for a dealer:)

I cannot reproduce in myself the results you describe. I optimize it on М1 (all ticks). History since 04.10.2004 (InterbankFX). Spread is 2 pips EURUSD.
At the best case it is about 0, but in other cases everything is falling apart, including those parameters that are shown in mql4.com picture.
Maybe I'm doing something wrong?
 
Indeed, this primitive behaves differently on the quotes of different brokerage companies.
If you are interested, I can send you the quotes that were used for testing.
sk@mail.dnepr.net
 
Indeed, this primitive behaves differently on quotes of different brokerage companies. <br / translate="no"> If you are interested, I can send you the quotes that were used for testing.
sk@mail.dnepr.net

If you have tested the analyzer with a different brokerage company and have not got it, please do not hesitate to contact us.
1...8910111213141516171819202122...309
New comment
Reason: