Once again, it's about the eternal: trend/flat.
It is very important for any strategy to identify the trend/float in a timely manner. One way or another, the effectiveness of the TS depends on it.
It is clear that on different TFs at the same point in time there may be both in the sense that a particular EA writer puts it, but let's try to talk more specifically - how to identify the f/o on the current TF? Who defines t/f? Or many do not bother to do it?
I approach the notion of flat as an arrangement of candlesticks one after another in a horizontal channel, it is very abstract, because many questions arise such as "how much should candlesticks fill the channel to be able to talk about a flat" etc.
At the moment I do it elementary - timeframes, say, 00:00-8:00 - flat, 08:00-22:00 - trend movements (one continuous directed movement or with changes of direction several times), 22:00-00:00 - flat. But this simplified method is very approximate, even though it improves indices of TS, but does not allow to use it on TF older than H1.
I am also thinking of playing with BB (Bollinger) to determine the TF, but do not know how to formalize it.
Please speak up.
My investigations on this subject lead to the idea that the flat should be characterized by its height (in pips) and length (in candlesticks). In principle it does not really matter when it happens. The daily movement over the last N candlesticks can be used as the starting point for measurements by height, by length - depending on the TF.
The characteristic of a flat, i.e. the determination of a flat is a flat, besides the suitable length and height should take into account, as you said, "to what extent the candlesticks should fill the channel" - so a flat can be characterized by three fractals, two of which are on one level (2+1) (at least). Ideally, 4 fractals (2+2).
My research on this has led me to believe that a flat should be characterised in terms of height (in pips) and length (in candles). In principle, it does not really matter when it happens. The daily movement over the last N candlesticks can be used as the starting point for measurements by height, by length - depending on the TF.
The characteristic of a flat, i.e. the determination of a flat is a flat, besides the suitable length and height should take into account "how much the candlesticks should fill the channel", as you said - so a flat can be characterized by three fractals, two of which are on the same level.
It's a good idea, we can calculate how much the neighbouring upper fractals differ from the spread between the last upper to the last lower fractal, so we can get t/f boundary conditions and even use fusilogy.
1. It is not actually necessary to trade at night. And with little volatility, it is also dangerous.
2. Quite right, it is also important to consider the length of the channel - I count it in extremums - the longer it is, the more favorable conditions for breaking out of it, simply speaking, a bounce from its border, but on the other side. I also consider the acceleration of volatility. But it works in the opposite direction - so I use the equilibrium of these factors.
What is a flat and a trend - I do not understand and do not make separate conditions for either because they are clearly seen afterwards, but it is also clear that they occur rarely in their pure form and one cannot build a system on them statistically - my tests have actually confirmed this. It is the human eye that automatically distinguishes trend-flat figures. The machine does it much worse, and there is no sense, because one does not wave one's fists after a fight...
I do not understand what is a flat and a trend and I do not make separate conditions for either of them, because you can see them afterwards, but you can also see that they are rare in their pure form and you cannot build a system on them statistically, which was confirmed by my checks. It is the human eye that automatically distinguishes trend-flat figures. The machine does it much worse, and there is no sense in it.
But even the simplest time filters suggest otherwise - strategies not only can be oriented to t/f, but should be.
The matter is that if we have determined that it is flat at the moment, we should simply wait for the positive equity and close the position. Logic can be quite different in one EA but at different times of the day, but if you have a numerical formula on hand, even a simple one, you can get even better results.
And what is a flat and a trend - I don't understand and do not make separate conditions for either of them, because they are clearly seen afterwards, but it is also clear that they are rare in their pure form and you cannot build a system on them statistically, which is exactly what my checks have confirmed. It is the human eye that automatically distinguishes trend-flat figures. The machine does it much worse, and there is no sense, because one cannot wave one's fists after a fight.
But even the simplest temporal filters suggest otherwise - strategies not only can be oriented towards t/f, but should be.
The matter is that if we have determined that it is flat at the moment, we should wait for the positive equity and close the position. The logics may be quite different in one EA but at different times of the day, but if you have a numerical formula on hand, even a very simple one, you can get even better results.
These are all myths; you can never determine the duration of the flat in advance. The most effective formula is the probability of an exit from the channel - the more time in the channel - the higher, the lower the volatility (volatility acceleration) - the higher.
As for the timeframes, I use two different ones for calculating extrema, and their exact value is determined by optimization.
These are all myths - you can never determine the duration of a flat in advance. The most effective formula is the probability of a situation of a channel exit - the longer the time in the channel - the higher, the lower the volatility (volatility acceleration) - the higher.
As for the timeframes, I use two different ones for calculating extrema, and their exact value is determined by optimization.
Andrey Dik:
Хорошая мысль, можно посчитать, насколько процентов отличаются соседние верхние фракталы от размаха между последними верхним до последнего нижнего, так можно получить граничные условия т/ф и даже использовать фузилогику.
If the inter-fractal space is low-volatile, it will be one of the signs of a flat.
Some fractal-forming candlesticks touch each other by ranges, i.e. there is no movement dynamics; fractals formed by false breakdowns (sometimes counter-fractals) do not show dynamics either, and the distance between the fractals may be "significant".
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
You agree to website policy and terms of use
It is very important for any strategy to identify the trend/float in a timely manner. One way or another, the effectiveness of the TS depends on it.
It is clear that on different TFs at the same point in time there may be both in the sense that a particular EA writer puts it, but let's try to talk more specifically - how to identify the f/o on the current TF? Who defines t/f? Or many don't bother to do it?
I approach the notion of flat as an arrangement of candlesticks one after another in a horizontal channel, it is very abstract, because many questions arise such as "to what extent the candlesticks should fill the channel to be able to talk about a flat" etc.
At the moment I do it elementary - timeframes, say, 00:00-8:00 - flat, 08:00-22:00 - trend movements (one continuous directed movement or with changes of direction several times), 22:00-00:00 - flat. But this simplified method is very approximate, even though it improves indices of TS, but does not allow to use it on TF older than H1.
I am also thinking of playing with BB (Bollinger) to determine the TF, but do not know how to formalize it.
Please speak up.