Calculation of the slope angle of the trend line. - page 18

New comment
 
-Aleks-:

This is from the field of cognitive psychology - i.e. conditioned by human physiological factors, yes, there may be deviations, as everywhere else, but we are talking here about a visually comfortably perceived image. Accordingly, a person will for some reason or other change the scale ("constant" it is per unit time of situation assessment) only in case of entering the area of discomfort. What is this area (and) you should find out, for example, when I change the scale in case of a strong trend and the graph visually rests on the upper or lower part of the screen, and you personally, when do you change the scale?

And, I'm not saying that the idea is 100% working, I'm just developing my thought and ready to give it up if the hypothesis is experimentally disproved.

If only it were a "hypothesis". Otherwise - completely vague judgments.

I asked for an example.

Here, a 10 degree sloping trendline, passing through 1.1170 at 8:00 on 22.05.17, on the Eurodollar.

What point should it pass through, in exactly 24 hours, and WHY?

What does your hypothesis say about it?

 
George Merts:

If only it were a "hypothesis". It's a completely vague judgement.

I asked for an example.

Here, a 10 degree sloping trendline, passing through 1.1170 at 8:00 on 22.05.17, on the Eurodollar.

Which point should it pass through in exactly 24 hours and WHY?

What does your hypothesis say about it?


The answer to the "Why" question is key - because, that's the scale most "people with money" have. What it is - I do not know - that remains to be seen. Accordingly, I will not be able to give you exact numbers at this stage. Above I have written what it could be and why it is so.
 
-Aleks-:

The answer to the "Why" question is key - because, that's the scale most "money people" have. What it is - I don't know - that remains to be seen. Accordingly I will not be able to give you exact numbers at this stage. Above I have written what it could be and why it is so.

Hmmm... Then it's not clear what your suggestion is.

Are you proposing some new degree measure ? Instead of "price per unit of time" (how trend is actually measured) and instead of "1/360th of a circle"(geometric definition) you propose your new definition "a degree is 1/360th of a circle assuming our scale is equal to some unknown quantity".

And how is this definition better than the first two ? In my opinion, it simply introduces an additional useless entity.

What is the purpose of measuring the slope of trend lines anyway? Obviously, so that they can be compared to each other and predict future price movements. But you propose to control the scale indicator additionally - and what's the point if it only complicates the comparison and prediction ? If you can't even indicate where your trendline is pointing, through which points it will pass in 24 hours ?

This extra entity could be tolerated if it were known. But as I understand it, it is not.

 
George Merts:

Hmmm... Then it's not clear what your suggestion is.

Are you proposing some new degree measure ? Instead of "price per unit of time" (how trend is actually measured) and instead of "1/360th of a circle" (geometric definition) you propose your new definition "a degree is 1/360th of a circle assuming our scale is equal to some unknown quantity".

And how is this definition better than the first two ? In my opinion, it simply introduces an additional useless entity.

What is the purpose of measuring the slope of trend lines anyway? Obviously, so that they can be compared to each other and predict future price movements. But you propose to control the scale indicator additionally - and what's the point if it only complicates the comparison and prediction ? If you can't even indicate where your trendline is pointing, through which points it will pass in 24 hours ?

This extra entity could be tolerated if it were known. But as I understand it, it is not.

Bury the unborn baby.

Any idea is subject to verification. I stated the idea and justified it, then you can either start testing it or reject it for a number of reasons, while you start making judgments about the results (effectiveness) of the proposed method without testing it.

My idea is that a person makes trade decisions and evaluates the market situation visually with the help of a graph, respectively the construction of a graph will influence the decision making regardless of natural indicators, i.e. if the market has suddenly started to move, it may be a strong trend for a person, while for the average value for N time period it will not be a big movement, but it will not be perceived as such because of the limited period of the visual series (i.e. if the average value is taken for a much longer period). In general, my method proposes to use not only mathematics but also human cognitive psychology.


 
-Aleks-:

In general, my method proposes to use not only mathematics, but also human cognitive psychology.

I don't mind, use it. I asked the question above. Take a trendline, which is inclined by 10 degrees according to you, and passes through a certain point. The question is which point this trendline will pass through in 24 hours, and why.

I am sure that if you really start getting formulas instead of saying general phrases - you will come to the same unit "price per unit time" just with some correction factor, which characterizes the "right scale of the chart". But only that - it is more reasonable to start from the price per unit of time, because the fewer entities - the more stable the system is.

 
George Merts:

I don't mind, use it. I asked the question above. Take a trendline, which is inclined by 10 degrees according to you, and passes through a certain point. The question is which point this trendline will pass through in 24 hours, and why.

I've said before - it all depends on scale, so there is no solution to this problem in degrees without defining the scale.

George Merts:

I'm sure that if you actually start getting formulas instead of saying general phrases - you will come up with the same unit "price per unit of time" just with some correction factor that characterizes the "right scale of the graph". But just that - it's more sensible to go straight to price per unit of time, because the fewer entities - the more stable the system is.

I work with different prediction options, including using velocity (points/time), but it's not always the best option, so I look for other options.

And, again, it depends on how to use this angle - let's say we have a trend and we measure its strength (including the angle of correction relative to the trend) - one task, but finding the third point through two is a completely different task...

In general, I do not understand such unreasonable non-acceptance of the idea - I will not prove my rightness, as I have not conducted any research on this topic - perhaps in the future.


 
-Aleks-:

In general, I do not understand such unreasonable non-acceptance of idea - I will not prove my correctness, as I have not carried out researches on this theme - it is possible in the future.

Why "unreasonable" ?

My argument is that using "angle at a fixed scale" differs from "price movement per unit time" solely by an additional unknown correction factor. But to breed entities beyond what is necessary is not the correct method of research.

And what about using the angle - why are we measuring the trend? Just to find the third point in the future - the point above or below which a certain event will occur. Of course, this point can be found by setting the angle when controlling the scale. But why this additional variable - I can not understand. Either it really improves the system profitability (or at least the accuracy of trend detection), or it is not necessary. I do not see how it improves the accuracy of trend detection - the trend 0.0001 per hour does not differ at all from the trend "one four-digit point per one bar H1", although the value differs by four orders of magnitude. But in the case of decimals - one value is easily converted to another. Using an angular measure, on the other hand, introduces correction factors based on trigonometric functions, and it becomes much more difficult to estimate values "by eye".

I used to make a custom trendline indicator - and there the question of the angle came up. The customer could not understand why one and the same angle on different charts gave completely different results. At first they started to "fix the scale" but finally they decided to measure the price movement per time unit and it solved all the problems about the scale.

 
George Merts:

Why "unreasonable" ?

My argument is that using "angle at a fixed scale" differs from "price movement per unit time" solely by an additional unknown correction factor. But to breed entities beyond what is necessary is not the correct method of research.

And what about using the angle - why do we measure the trend? Just to find the third point in the future - the point above or below which a certain event will occur. Of course, this point can be found by setting the angle when controlling the scale. But why this additional variable - I can not understand. Either it really improves the system profitability (or at least the accuracy of trend detection), or it is not necessary. I do not see how it improves the accuracy of trend detection - the trend 0.0001 per hour does not differ at all from the trend "one four-digit point per one bar H1", although the value differs by four orders of magnitude. But in the case of decimals - one value is simply translated into another. Using an angular measure, on the other hand, introduces correction factors based on trigonometric functions, and it becomes much more difficult to estimate values "by eye".

Some time ago I was making a custom Trendline indicator - and there the question of the angle was raised. The customer could not understand why one and the same angle on different charts gave completely different results. At first we started to "fix the scale" but finally we decided to measure the price movement per time unit and it solved all the problems regarding the scale.

So, that's what I wrote, for the purpose you described - the point/bar method is good. I am talking about using the angle for the purpose of defining limits, i.e. it is clear that a 90 degree trend is an inadequate situation and will not last long, and a pleasant to the eye trend of X degree and the market will tend to it (also it is theoretical, because there was no test, but the justification is the same - human psychologists).

 
-Aleks-:

OK, so I wrote that for the purpose you described, the point/bar method is good. I'm talking about using the angle.

Well, that's what I want to see an example, which would demonstrate the advantage of such definition of the trend.

So that we can see - if we define a trend as points per bar, we get so-and-so and it's not good.

But in the same situation, if we define the trend in degrees at a certain scale, we get something else, and it is seen that it is much better.

  
double RadToAng(double drad)
  {
   return NormalizeDouble(drad*180/3.1415926535,4);
  }
//+------------------------------------------------------------------+
//|                                                                  |
//+------------------------------------------------------------------+
//+------------------------------------------------------------------+

int getAngle(double p,double price1,double price2)
  {

   double alfa=MathArctan(1/MathAbs((price1-price2)/p));
   return (int)(RadToAng(alfa));
  }
calculation of the angle in degrees, and how to get 2 points from the trend is explained more than once
1...111213141516171819202122232425...27
New comment