Channel, what are you? - page 2

 
Something else I'd like to say. It's no problem to build a canal. Anything.
What's important is what it's for. What to do with it. HOW to trade. It would be interesting to discuss the PRACTICES of working in the channel or with channels, whichever you like.
But in this case it is just another thread on the subject of "Have you ever seen such a drawing?
 
For almost an hour I have been looking at the graphs and trying to comprehend what I have seen and read in this thread. So far I have come to the following idea about the order of my own gray matter work:

1. At first my brain, analyzing graphic data (I skip the stage of transformation of light impulses on retina into meaningful quotations:), chooses a scale for analysis. I don't know about others, but personally mine is clinging at once to some parts from one tenth to one third of the current chart window size. Well, generally speaking, if we want to solve the problem of programmatically determining presence of a channel, we can solve the problem of scale selection directly - take sections from M to N bars and search there. If we repeat it several times, we will get a normal choice of scale, it's not critical for the complexity of the problem. Let me once again repeat this to keep the discussion on the same subject: we are solving the task of finding the already formed channels in the history.

2. My brain chooses from all sectors to be further analyzed those which are approximated rather well by a linear dependence on time. As I understand it mathematically, "reasonably well" can be expressed as follows:
a) we draw tangents to the chart section from above (by High) and from below (by Low) and from each side we choose the one with the minimum sum of squares of deviations from those High (Low), which are not directly touched by the tangent. Instead of the sum of squares we can use another function (for example, the sum of modules), I think the results will not differ much. There is one catch: even with this method the best tangent based on visual inspection may not coincide with the best calculated one. The reason is small random outliers that sometimes appear on the channel boundaries and slightly "stick out" beyond them. Moscow ignores these outliers, filters them out, as if they were absent, so the program should do the same. Considering that the algorithm for finding a tangent should check all possible pairs of points it can go through, we may consider not only "strict" tangents but also those which "leave out" small spikes - for example, not more than a percent of the mean deviation, e.g. not more than a given percent of the mean deviation of High/Low from the tangent.
b) Measure the angle between the upper and lower tangent and if its tangent does not exceed a pre-specified small value, assume that the tangents are "almost parallel". Besides, it is useful to analyze the angles between the tangents and the regression line for better reliability of the result.

3. And now the most complicated part. At "sufficiently linear" segments our brain plays the trick - the price should behave "sufficiently regularly" in the channel. Let me explain: if you look closely at the channels on history, you will notice that there is usually no chaotic price movement in them. On the contrary, in most cases more or less oscillatory movement from one wall to the other and back prevails. Therefore, I propose the following as the next necessary formal condition of "channelability": the square of the sample autocorrelation function of a price series on the analyzed segment should have at least one more expressed (e.g. not less than 0.5) lobe in addition to the main one.

Opinions please:) especially on the third point
 
Svinozavr >>:
Чтобы еще хотелось сказать. Канал построить - не проблема. Какой угодно.
Важно - для чего. Что с ним делать. КАК торговать. Вот ТАКТИКИ работы в канале или с каналами - как угодно - было бы интересно обсудить.
А так - еще одна ветка на тему "А такой рисунок видели?"

One tactic is to trade on a channel break. But in order to do that, the channel must first be determined. So personally, I do my favourite task - formalisation of difficult to formalise tasks - not for the sake of drawings, but for a very specific self-interest:)

 
alsu >>:

одна из тактик - торговля на пробой канала. Но для этого наличие канала надо сначала определить. Так что лично я занимаюсь своей любимой задачей - формализацией трудноформализуемых задач - не ради рисунков, а ради вполне конкретного корыстного интереса:)

When I look at channels, I can quite clearly divide them into two classes - conditionally "beautiful" and "ugly". The former, from my observation, get through much more often in a "pretty" way...

 
I've tried a lot (a lot - known and purely ogyginal) of different channels. The tactic of entering on a breakdown, with a competent construction of the channel (yes, I agree, how to build it is a topic) - is profitable. BUT. I have seen such profits "in a coffin and in white slippers". That is why I started to talk about tactics - there are different variants of processing of the same breakdown, refinements/filters of passes using additional (e.g. oscillators) indicators, different stop patterns, fixes and MM in general.
Therefore I would like to see proposals/discussions on this scheme:
- The principle of constructing the channel: what is the next rotten feature of fucking Forex, we are trying to catch.
- Tactics: HOW we use it.
- MM.

I.e. it seems to me that it is meaningless to tell about one or another variant of a channel without a way of its use. It would be - I repeat - "Funny pictures" & nothing more...
Without a strategy - zeros. Although, if you consider them as indicators - why not. Sapienti like sat. But you have to be a big sapienti and art lover for the sake of winning - pardon - for the sake of art, for that to be sat.
 
alsu >>:
уже почти час смотрю на графики и пытаюсь осмыслить увиденное и прочитанное в данной ветке. Пока пришел к следующей мысли о порядке работы собственного серого вещества:
...

Я бы все-таки, определился с темой: 1. Subj. или 2. Канал, каким он Вам представляется?
Вторая формулировка в большей мере соответствует Вашим рассуждениям, но я хотел бы, для начала, коснуться первой.

В самых различных процессах понятие "канал" возникает в случае, когда в силу недостаточной управляемости объекта, не удается сразу установить его движение строго по требуемой траектории.
Например, при заходе самолета на посадку. Болтание вправо-влево по курсу и вверх-вниз по глиссаде и образует соответствующий канал. Границы такого канала должны сходиться в заданную точку, куда хотелось-бы приземлиться (консолидация, однако).

Ну, а ближе к теме,- канал в нашем случае образуется вследствие инертности рынка при отсутствии новостей в период выхода на уровень/границу поддержки/сопротивления. Народ знает, что по правилу двух КУ надо то-ли закрыться, то-ли открыться, если что-то при делении на что-то другое похоже на 1.618 или 0.618,- вот и нервничает, причем: каждый по-своему. Вот Вам инерция. А новостей нет,- бежать, вроде некуда. Вот и начинается: Breakdown - Recoil ... . Я обычно такие штуки рисую в виде границ, параллельных оси канала (но не симметрично), по "подтвержденному" удалению от оной. В качестве оси выступает уровень/граница поддержки/сопротивления, либо тренд при его пробое. В качестве сигнала подтверждения используется факт пробоя осевой после прохода нового экстремума относительно осевой.
 
alsu писал(а) >>

When I look at channels, I can quite clearly divide them into two classes - conditionally "beautiful" and "ugly". The former, from my observation, get through much more often in the "pretty" way...

>> there are a lot of questions about this subject which you can answer if you want to build a super profitable system:
1. why the initially built channel changes
2. why the price breaks through it and then comes back to the channel
3. why the channel widens
4. when a breakout of the channel is possible.
5. when having broken the channel, the price forms a new channel
6.how long will the price stay in the channel?
By answering these questions, you can understand how the market behaves, know how the price behaves, when to buy or sell, and on the other hand, you can enter or exit the market with accuracy.
You do not need to ask me counter-questions, I have already found all the answers, and whether you will do it is your life and your success...

 

A channel can be defined as a range formed by a support and resistance level whose position varies uniformly over time (sloping channel) and is constant (horizontal).
Any support and resistance is a concentration of pent-up supply and demand. Major buying and selling that prevents the price from moving out of the range. Therefore determining the limits of the channel is all about determining levels of concentration of pending orders.
The logic of horizontal channel formation is a bit easier, but for the sloping one we need to find out what makes this range change uniformly for some time? Perhaps, this is the market maker on a low timeframe, trading some analog of the spread, or the fixing of profits and losses on higher timeframes based on a certain yield changing evenly over time (depending on the swap, for example). The case is dark with these sloping channels in this formulation.
But we may define the channel differently - by only one median line, and visually finding inside the channel - by attracting to it. Projected onto the trading logic, the median price could be some analogue of a fair price, or the opinion of participants about it. In this case, linear regression is useful in constructing it. The width of the channel depends on the current volatility.

 
Avals >>:

С логикой образования горизонтального канала несколько проще, то для наклонного нужно разобраться - в результате чего некоторое время равномерно изменяется этот диапазон?


Is it possible that a channel is formed based on participants' expectations that "it looks like a channel is emerging"? That is, a formation resembling a channel emerges in random rambling which is then asserted as market participants begin to employ channel tactics?

 
neoclassic писал(а) >>

Is it possible that a channel is formed based on participants' expectations that "it looks like a channel is emerging"? That is, a formation resembling a channel emerges in random rambling which is then asserted as market participants begin to employ channel tactics?

imha it is possible but closer to a channel with median price logic. Because one line is simpler and more generalised than two, and even with a fixed distance. This channel should be seen by many people at once, while for the median channel, only the current slope should be seen, and its width is determined by volatility.
In any case, the condition of formation will be that the market has a large number of participants of approximately the same trading horizon. Then the amount of history that they analyze is almost the same, and the channels built on the same piece of history will be very close. If we add 10 traders and give them task to draw one channel for each intraday, they will draw almost the same thing, comparing to giving them more freedom in choosing time frame for the channel. In that case, everyone will draw his own channel and trade it. There is no self-fulfilling prophecy then.