Discussion of article "How Long Is the Trend?"
link error
1.3. Третьим был рассмотрен Percentage of Trendthe link https://www.mql5.com/en/code/477 from point 1.2 is given.
should be https://www.mql5.com/en/code/868
- votes: 15
- 2011.08.24
- Nikolay Kositsin
- www.mql5.com
Drawing criticism.
My assessment of the article:
-- for effort -- GREAT
-- for content and disclosure of the topic -- NOT satisfactory
Recently there was an article of the same authorship "Comparative Analysis of 10 Trending Strategies" https://www.mql5.com/en/articles/3074.
How can one examine trend and/or flat, draw conclusions and not define trend and flat?
On what basis does the author conclude that the indicator he has chosen is a trend indicator? Where are the references to the author's conclusion on the correct use of the indicator?
For example, the author of the article writes:"1.1. Let's start the study with the classical trend strength indicator ADX. The TrendLevel will serve as an estimation of trend or flatness . We consider that there is a trend if the main line rises above this level."
Open, so as not to search too long, the most accessible source https://ru.wikipedia.org/wiki/Система_направленного_движения and read:
Wilder believed that rising and high values of ADX and ADXR indicate a strong trend, while falling and low values indicate a trendless movement[3]. The actual direction of the trend can be judged by the ratio of +DI and -DI.
With this in mind, we can suggest, for example, the following trading strategy[3]:
- For long positions:
- Buy if +DI > -DI and ADX is rising.
- Sell if +DI < -DI or ADX is falling.
- For short positions:
- Sell short if +DI < -DI and ADX is rising.
- Close the short position if +DI > -DI or ADX is falling.
-- and a link is provided to a respected source: Colby Robert. Encyclopaedia of technical indicators of the market. - 2nd edition - M.: "Alpina Business Books", 2004.
I.e. the author of the article took an indicator, applied to it unexplainable rules -- unsubstantiatedly called these rules "trend identification" -- and made absolutely left conclusions based on the left rules.
Perhaps, this is the first article that attempts to study market changes retrospectively, which is good.
However, the sample of instruments is small, and besides the analysis is made on weekly bars, it is much more interesting to look at the change of trends on small TFs - for example, hours or fifteen minutes. During a week the market can go both ways more than once on small TFs.
My own observations say that there are no changes on small TFs.
Thank you for the interesting article.
The trouble is that traditional methods of technical and wave analysis (and, accordingly, indicators based on them) do not have a precise, unambiguous definition of the concepts of "trend" and "flat".
Therefore, until now every trader and analyst defines the start and end points of trend and flat in their own way.
In addition:
1. Algorithms of traditional indicators have an arbitrary search depth for sampling historical data (for whom it is convenient).
2. The algorithms of data sampling processing do not take into account the nature of financial markets, namely: they do not take into account that
not only the amplitude, but also the frequency of fluctuations in the price of financial instruments is constantly changing.
Do the above-mentioned indicators take into account the factor of constantly changing frequency of market fluctuations?
Certainly, no.
That's why Alexander's researches are interesting as always, but they are researches within the framework of absolutely outdated analysis methods.
methods of analysis. And this is a global problem, not the author's mistake.
I would even say that the concept of trend and flat is a borderline concept, and this borderline is so blurred that one has to start from certain axioms when researching. In this article I confirmed for myself the only thing, probably not very pleasant for traders, that markets have become even more difficult to predict due to the fact that there are many times more factors influencing them than 5, 10, 20 years ago.
Imho, if trend and flat are borderline concepts, then I am a Spanish pilot. Another thing is that there are phases where one transitions into the other. Then uncertainty increases with all the consequences.
Imho, it is conceptually important to give them clear, algorithmic definitions when describing trend and flat. Even if they are the simplest, and I am not afraid to say cheesy. The article has it to some extent: the author uses various indicators. But it would be worthwhile to start with the simplest setups. For example, consider a series of bars with increasing lows as a bullish trend. And play with the size of this series. And pull in the distributions for the series. Well, then statistics can help. And then it is almost a full-fledged study.
I've looked at the code, in particular the"TrendCount" Expert Advisor. No comments. And classes and procedural programming... all in a heap...
And in general, the author has attempted to tackle a deep topic, for which he has a lot of respect.
I would even say that the concept of trend and flat are borderline concepts, and this borderline is so blurred, ....
If trend and flat are blurred concepts (in your opinion).
I.e. you cannot clearly identify trend and flat (on history).
And your article is called "How long does trend/flat last in % of time".
What have you researched then? And how adequate are your conclusions?
... When researching, one has to start from certain axioms. ...
What axioms did you start from in the article? If you take an obscure indicator and apply an obscure interpretation to it (actually, this is what is done in the article), you will get only one axiom: "The grandmother said two words".
What axioms did you base your article on? If you take an obscure indicator and apply an obscure interpretation to it (actually, this is what is done in the article), then you will get only one axiom: "The grandmother said two words".
I had an idea to use one of your Blau, then "unclear which" probably wouldn't be, wouldn't it Andrew?) I've already got your Neud. Although no one has received a UD as far as I know. So...so.
I agree with this statement:
The trouble is that traditional methods of technical and wave analysis (and accordingly indicators based on them) do not have a precise, unambiguous definition of the concepts of "trend" and "flat".
Therefore, no one can clearly and truly identify trend and flat for everyone. And it turns out that no matter how anyone researches, there will be a conditional Zelinsky with the statement "Grandmother said two words" and go to the garden.
I had the idea of using one of your Blau's, then the "obscure one" probably wouldn't be there, would it Andrew). I've already got your Uncle. Although no one has received a UD as far as I know. So...so.
I agree with that statement:
Therefore, no one can clearly and truly identify trend and flat for everyone. And it turns out that no matter how one researches, there will be a conditional Zelinsky with the statement "Grandmother said two words and go to the garden".
Have you tried to search for "forex trend" -- and easily see the answer to the two key questions of your article: 1) what is a trend/fleet and 2) determining whether or not it is trending/fleeting?
I think you will be surprised by the answer, because trend/flat is definable and there are quite clear definitions for it.
There is a problem of trend/flat forecasting -- but there are no problems with trend identification in history and never have been.
... I've already got your "B". Although no one has received an OD, as far as I know. So...so. ... ...
In my practical experience -- there are only two grades -- OTL and NEUD (as in not OTL).
I.e., anything that is not OTL -- then it is unequivocally only a NEED -- just like you can't be a little bit pregnant.
Even in childhood one learns a poem by V. Mayakovsky: "What is good and what is bad" http://lukoshko.net/poetry/poetvm1.shtml:
The boy
joyful went,
and the crumb decided:
"I will
do good
and I will not
- bad".
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
You agree to website policy and terms of use
New article How Long Is the Trend? has been published:
The article highlights several methods for trend identification aiming to determine the trend duration relative to the flat market. In theory, the trend to flat rate is considered to be 30% to 70%. This is what we'll be checking.
1.1. We will start our research with the classic trend power indicator ADX. We will evaluate the existence of trend or flat using the level of TrendLevel. Let's assume that the trend exists if the main line rises above this level. Fig. 1 shows an example of determining Trend Zone and Flat Zone using this method. The market state will be calculated for the number of candlesticks, on which the ADX value is higher than TrendLevel from the given sample.
Fig.1 Using ADX to determine the trend/flat zone
Author: Alexander Fedosov