Looking for patterns - page 80
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
Now about the Genghis indicator. I got some figures based on it, which I will give in the table below. The table shows how the time of price movement in the trend changes depending on the price overcoming the minimum distance that we set in the input parameters of the indicator. The study was conducted on the EURUSD chart on 15 minutes for greater accuracy, from 2000 to 2020. All points are in 4 digits.
So far I have only studied the time, later I will also study the distances covered by these trends.
One wonders whether in this case the square root law (SQL) underlying the formulas for the spread (called variance) in Alexander_K2 works?
Very interesting, is the square root law (SRC) underlying formulas for variance (called dispersion) in Alexander_K2 working in this case?
Alas, not interesting, Vladimir...
The task is set quite clearly - to find, identify areas or clusters in the time series, for which the ZKC works. If the trader initially doesn't set this task, but simply transforms the series, plays with it, then you don't need to look - nothing will work.
As Alexey correctly pointed out yesterday, so that the figures obtained do not remain empty sounding, I will try to start interpreting them.
The Chingiz indicator has a simple idea that if the price has gone from the low to the current moment, for example, more than the value we set, then we register an up trend. Then, if the price does not go down from the new high below the set value, we consider that the upward trend continues. Thus, we break up the entire chart into sequentially alternating trends. The trends are measured from one extreme to the other. It is natural that trend registration happens later with a delay when the price has gone part of the way but it does not affect the accuracy of measurement when studying the history.
We considered the period from 03.01.2000 to 28.02.2020. In this period we got 5202 working days. If we consider the first row from the table, where we have chosen the minimum distance for trend fixation to be 20 points, we get 62200 trends for the whole period under study. That makes almost 12 trends per day. But visually we can see that 1-3 significant movements occur during the day or not occur at all. This may mean that 12 trends per day is noise.
Doubling the minimum distance to 40 points gives a reduction in the number of trends by almost 4 times. We are now registering 3 trends per day.
As a result, to get closer to a visual assessment of what is happening we need to set 60-100 points when we can record one trend change per day. This is what the median time shows. The median time is overestimated due to large values, which are extremely rare. Generally, the frequency of repetitions is squeezed closer to the minimum values.
As Alexey correctly pointed out yesterday, so that the figures obtained do not remain an empty sound, I will try to start interpreting them.
....
The bottom line is that to get close to a visual assessment of what's going on, we need to set 60-100 pips, when we can record one trend change per day. Which is what the median time shows. The median time is overestimated due to large values, which are extremely rare. In general, the frequency of repetitions is closer to the minimum values.
You did not take into account the peculiarities of the market. You took a period in which there were daily fluctuations of 200 pips and there were years in which the daily fluctuations did not reach 100.
Such averages are extremely negative for further forecasts.
Well the average is a fluctuation boundary and you can use it, but deviations from it are not constant. Bolinger's channel shows it very well. You don't have to invent anything. A ready-made solution for your study.
So many attempts have been made, I don't see the "rest stop" at all...
I've gone into teaching. There's more money there :)
Uladzimir, yes you can do a breakdown by year and compare with the total. See how the values change. That was the idea.
And I'm suspicious of all sorts of averaging strips. Point to point only.
There is a small request to delete most of the text when replying, so as not to clutter it up.
Uladzimir, yes you can do a breakdown by year and compare with the total. See how the values change. That was the idea.
And I'm suspicious of all sorts of averaging strips. Point to point only.
There is a small request when answering remove most of the text, so as not to clutter it up.
(I have deleted unnecessary text).
The market has too many averages to understand where the average price is.) Only from point to point is more reliable.
Thank you, Uladzimir! That's right!
Now by the numbers so far the obvious things are coming out. I think by investigating changes in several parameters at the same time, one can find interesting features. But not everything comes out quickly. Time is always short.
Thank you, Uladzimir! That's for sure!
Well, the numbers so far have been making the obvious points. I think when investigating changes in several parameters at the same time, you can find interesting features. But not everything comes out quickly. There is always a shortage of time.
Time flies away instantly in my coding. A week flew by like a day.
"when examining changes in several parameters simultaneously..." I don't get which ones?
When several conditions are superimposed, for example, how long the trends lasted after the opposite trend of 500 pips. There can be many such conditions, and interesting findings are possible.