![MQL5 - Language of trade strategies built-in the MetaTrader 5 client terminal](https://c.mql5.com/i/registerlandings/logo-2.png)
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
Looking for ideas on how to automate all this.
It is trivial on the history. You draw a channel so that price touches each of its borders three times and mark points at the points where the price touches them. The real time is problematic: the channel is not yet drawn and it is still unknown where and when the price is going to kiss it.
Already explained on the first page how to build a canal before it is formed point by point. Has something been misunderstood or forgotten?
Already explained on the first page how to build a canal before it is formed point by point. Have you forgotten or misunderstood something?
I'm not arguing, I'm saying the same thing. Compare them:
gpwr:
...
5. A channel usually lasts until price touches one of its boundaries 3 times.
...
Reshetov:
On the history it is trivial. Draw a channel so that price touches each of its borders three times and mark points at each touch point.
...
So here's a question for you: how do you know in the present where the price will touch the channel boundaries in the future?
You have confused the cause, i.e. the price with the consequence, i.e. the channel. The channel, as any other markup, is constructed based on the price, not the price on the channel.
If it were the other way round, everyone would be rich and happy. No one would be working, and everyone would only be drawing channels in the directions the price would need to move to the tees.
Vladimir, if you don't answer them, they will go away on their own.
Develop the topic, it's interesting. And just ignore the idle talk.
So here's a question for you: how do you know in the present where price will touch the channel boundaries in the future?
We do not predict touch points. We predict the channel lines and that is the art. The important thing is that the predicted channel should be about the same width as the past ones. Again, it is all written on the first page. We predict the channel, wait for touch points and trade on reflections from the trend line. Of course, the initially drawn lines will correct, but not by strong price deviations from the channel boundary, which allows profitable trading with small stops. Point 2 (the second touch of the upper down-trend line or the second touch of the lower up-trend line) is the most profitable. Haven't you ever read about trading in channels or along support/resistance lines? Same logic. No one there predicts when the price touches those levels. And when it touches those levels, they trade either reflecting them, or breaking them through. I'm embarrassed to explain all this to the "experienced".
That's all, I am not replying to your further messages, as Andriy advises. You have made two pages of nonsense.
The important thing is that the predicted channel should be about the same width as the past ones.
I don't know how it was "in the stock market 6 years ago", but in forex for the last 15 years the channels are about the same width as the past ones - not very often
If you want to check - I can write the code on a commercial basis
I don't know how it was "in the stock market 6 years ago", but in forex the last 15 years the channels are about the same width as in the past - not very often
If you want to check - I can write the code on a commercial basis
No thanks. Your code will draw channels of different widths and I don't need them. Understand that you can draw a bunch of different channels of different width on one and the same part of history and then talk about changing volatility, etc. The commonly used method to measure volatility is to subtract a mask from a price series and measure the residual RMS. If we use this method, we will find that the RMS varies over time, which means that the residual is non-stationary. But this conclusion is only true if we take away the mask. Why not take away any arbitrary curve? For example, I can choose such a curve that the residual will have a constant RMS. This is the essence of my strategy. I want to fit such a curve that the price has approximately the same maximum deviation from this curve. Mathematically, this problem is reduced to finding the best fit curve so that the error (price deviation) has certain statistical properties (stationarity, for example). An empirical solution to this problem is a broken straight line. Moreover, depending on where the breaks are, we may get different maximal deviations of the residual from this broken line, which leads to the existence of many solutions-channels of different widths. I have chosen a width that meets my requirements and am trying to automate the process of finding channels of a given width. My substantiation for choosing a constant width is that the middle of the channel is essentially a trend line. The maximum deviation from the trend lines should be approximately the same in different parts of history, so we have the same players with their strategies, emotions, etc. I don't want to claim that this strategy is correct and I won't argue about it.
No thanks. Your code will draw channels of different width, and I don't need it. Understand that you can draw a bunch of different channels of different width on one and the same plot of history and then argue about the volatility change, etc. The commonly used method to measure volatility is to subtract a mask from a price series and measure the residual RMS. If we use this method, we will find that the RMS varies over time, which means that the residual is non-stationary. But this conclusion is only true if you subtract the mask. Why not subtract any arbitrary curve?
Because the variance and RMS are by definition calculated from the arithmetic mean.
For example, I can choose such a curve that the residual will have a constant RMS. This is the essence of my strategy. I want to fit such a curve that the price has approximately the same maximum deviation from this curve. Mathematically, this problem is reduced to finding the best fit curve so that the error (price deviation) has certain statistical properties (stationarity, for example). An empirical solution to this problem is a broken straight line. Moreover, depending on where the breaks are, we may get different maximal deviations of the residual from this broken line, which leads to the existence of many solutions-channels of different widths. I have chosen a width that meets my requirements and am trying to automate the process of finding channels of a given width. My substantiation for choosing a constant width is that the middle of the channel is essentially a trend line. The maximum deviation from the trend lines should be approximately the same in different parts of history, so we have the same players with their strategies, emotions, etc. I don't want to argue that such a strategy is correct and I won't argue about it.
What you start subtracting residuals from other curves will not make stationarity appear. What will emerge is a fit to some piece of historical data - econometric casuistry.
Breakdown of the previous channel. EURUSD should reach 1.300. Judging by the pattern, we are likely to expect a flat (horizontal channel). When we see a local bend downwards, we will be able to say more accurately where the new channel will be. It is dangerous to trade at the moment. If the price bends down and then rebounds from the upper boundary of the previous channel, it will confirm the up-trend.