First sacred cow: "If the trend started, it will continue" - page 4

 

Here's a real-life example ...

you can see on the chart that the price is making an upward move...

but in fact the main direction of movement is still downwards...

In the first case, the chart was shown on M1, in the second the same area only on M5

the price just bounces from the border to the border of the channel,

if we add up these waves we will get the same fluctuations as we can actually see on the price chart ...

So what Neutron wrote about correctly and reliably shows what happens to price ...

But using trend indicators and oscillators we see a narrow-directional pattern, so we perceive any movement as a reversal,

that's why we always want to see a unidirectional movement without delays, without short-term reversals...

 
Mathemat писал(а) >>

Hi all.

So, three concepts: "trend", "started", "continued". How do you understand them in the context of the branch's topic?

Note, by the way, that this maxim does not work when applied to an ordinary Wiener process. Yes, in a Wiener process there are areas that can be called local trends, but, alas, in the long term they can not make money on them. And the statement itself actually confirms the possibility to earn on a trend!

Hi Alexey ! If there were an established, correctly defined notion of a "trend", the mathematicians (excuse me, it's not a pun :-) would have gutted it out. Since there is no such concept, there are two options: 1. to discuss here who thinks what about it - it is interesting, but hardly constructive; 2. to give one's own definition of the concept, which would limit it, but make it mathematically clear - then we may study statistics of the market based on it and (at least) find out if it exists or not. And if it turns out that it does, then questions can be raised about when, how and for how long.

From my point of view, a trend can be defined either as a segment of zigzag with height/length/ ratio not less than or a channel with sigma/length ratio not less than.

If there is a trend, the most appropriate model for it is the Bernoulli process with unequal p,q. Here you can see, for example, what shape the distribution of sigma/length ratios has for Bernoulli and for the real market. Or how often trends with a given ratio occur there and there.

In general, the correct question depends on the chosen trend definition and on what we actually want to know.

As far as I understand, the matter in question is not how to find out whether the trend starts, but whether it exists at all. I think this question may be answered.

Neutron wrote >>

Take another look at the figure. You can clearly see that your statement is wrong:

Price moving in general direction, on different TFs (in this case t1,t2 and t3) moves in different directions!

Hi Sergei ! Your model is quite convincing. There is only one detail. It is stationary. And we can't say that about the market. We cannot say the opposite either. That's why I personally see the question as the question of local behavior and its model. If periodically there are sections in price behaviour that are described by a model, and if this model by local parameters gives the possibility to predict the total length of a trend or the moment of its end, what difference does it make if it is all in all a superposition of flats or not?
 
Neutron писал(а) >>

You look at the figure again. You can clearly see that your statement is wrong:

Price moving in general direction, on different TFs (in this case t1,t2 and t3) moves in different directions!

Just as there is no trend so there is no flat. This is a history, a coincidence. The price does not remember all this wave structure, and neither does it have a correct frequency decomposition. When it occurs, what it manifests in and how long it lasts.

 
Avals писал(а) >>

Just as there is no trend, there is no flat. It is a story, a coincidence. The price does not remember this whole wave structure, nor does it have a correct frequency decomposition. When it occurs, what it manifests in and how long it lasts.

Look at the market without all that knowledge and experience, as if you were born and saw a curve...What it will represent (not remind) to you...Then tell me what the market is...While there is knowledge and experience and past days of intensive work( experience), in all manifestations the trader will always see figures, inclinations, waves...and attempts to bind everything he sees again to the past experience...To create something new you have to abandon the old - this is the real reality of progress (plus lazy automation, which is also necessary).

 
Yurixx >> :

Hi Sergey !

>> Hi, Jura!

Glad you took part in the discussion and of course I agree with your point of view on the matter. Of course, the topic raised by Mathemat-om is provocative and more of a joke. Indeed, it has been discussed many times and I will reiterate that the most correct definition of market conditions is given in Pastukhov's work and it is determined by the difference between the H-volatility and the equilibrium 2H level (it is an indicator of trendiness/flatness). This estimate is a stationary characteristic of a particular instrument and for the vast majority it is less than 2H? It is for this reason that I speak of the provocative nature of the author's entry in this thread.

The trend would rather reverse than continue! This is the law for market prices.

 

In my opinion the notion of Trend is a relative concept as well as almost everything else. In order to determine (at least to try) if a "trend" will continue or not, one would have to identify it, which can easily be done, but only on history ))

 

Does it also need a story to tell that the trend is going on and when it will end...?

Or add more pictures?

 

In this case it is not difficult to understand that the trend is bearish. But it only applies to this TF. On H4 it might be bullish.

The difficulty is in determining the end of the trend. From this picture it is possible to make some assumptions about the further trend movement (but for me it is 50/50), but not more.

You can determine the trend target with 1p accuracy only when there will be a reversal... ))

Of course, you cannot do without images. If you have pictures from the future, we will be glad to see them...

 

The thing is that it's hard to tell which trend in this case will be difficult to look at different TFs...but if you play in the direction of a larger movement then it turns out they will all be unambiguously either ascending or descending...

The question on the trend is always to determine that the future trend will definitely move in such and such a direction for so many points...

But even the above picture shows that the moment comes when we can say that the trend began to reverse its movement in the opposite direction... It seems to be true, but in order to say with certainty that this is the case, you need to look at exactly the same picture but already on a older TF and then the question of what happens - will definitely disappear ...

nobody will ever tell you about the reversal moments with 1p accuracy - in this case you need more extended information with the frequency of a 20-second TF...then we could say that the price will turn, but even with such a turn there is an additional triggering threshold within 2-3 points.

The sense of it all can be summed up in a couple of lines:

We are trying to play with bulls and bears trends which show maximum NN pips, with a trigger threshold of 4-7 pips.

And as for the other pictures - so far there is no area to show that at the moment the trend will move in different directions on different TFs... About what is going on at the moment you can look here and on the next page... on page 62 was the beginning of the forecast, which coincided exactly, only with a full analysis of four TFs...

 
forte928 >> :

It seems to be true, but in order to say with certainty that this is the case, you need to look at exactly the same picture, but on an older TF and then the question of what is going on - will definitely fall away...

Who told you that it will disappear? What can you be sure to find out with the help of older TF?

IMHO, what helps in this case (and to a very small extent, too) is the wave analysis, and it is on the lower TF. Then we will know if it is, say, the fifth wave has ended or only the first one...