a trading strategy based on Elliott Wave Theory - page 309

[Flash555]

Денис Панкратов:
very useful thing wave analysis))))

I will write a separate post about wave analysis. By the way-wave analysis is only a special case of trading on patterns (IMHO), and it is more logical to talk about a wide sense of the word, rather than exclusively about waves. Because the configurations of waves are not unambiguous either.

There is an opinion that the analysis of patterns is (if I may say so) analogous to the degree of fractality of a series, or to the dimension of fractality.

Which should change the trend more smoothly. I have some ideas in what direction to look, but I obviously do not have brains enough to codify it.

Because volumes are huge in view of variety of floating conditions and floating lengths of samples on which these conditions will be

to be identified.

1-The first stone is that the pattern can have completely different dimensions in both axes.

Besides, a pattern is a condition, and you can set a lot of conditions for each individual study.

And the longer a pattern is (the number of pattern points or the number of degrees of freedom), the faster the power consumption for its checking grows.

2- The same pattern does not have to go strictly consecutively to the end of its extreme counterpart,

It may go with gaps. Or it may even be able to cover the last pattern in broad strokes. Just like a trend with a flat.

That is why it is a problem to identify them. There are sectors that belong to both the trend and the flat.

Both problems can be solved only within the framework of a certain task.

In general, the idea was to search for each pattern on the minimum timeframe.

For example, we take a certain pattern of 5 points - Gartley's butterfly. And we search from the edge into the history and find the first one,

then from the edge of the first one we start searching for the second one and so on. Then each pattern will be a new row (some kind of

of a non-linear TF) where we will look for the same pattern by the same conditions. And so on, increasing the abstract octave.

Besides, we can simultaneously calculate the statistical significance of patterns on history (but I think it's not necessary in this calculation method).

As a result we take the first pattern of the lowest range, say it consists of 20 bars. Then we take the first pattern of high - the way it consists of 25

Then we take the first pattern of the lowest one (nominally 25 bars) and the highest one, etc. We can see that they will have a similar look if we compare them side by side.

But they will consist of a different number of their own "bars" and different amplitudes analogous to fractals of different order.

It seems this information should be taken into account when choosing an initial pattern to be checked. Here or on the history

and see how the characteristics of pattern working vary at different levels of fractality depending on the changes of the conditions of the pattern itself.

Or we can compare the change of sizes by the number of bars and the amplitudes at different fractality levels. It is possible to identify the pattern on the higher tf

even before it was formed through information about how it was formed in lower bits.

Or something else, we can think in more detail if we want. It seems to me that it is well known patterns like head-shoulders or triangles,

or even waves, would seem to be statistically significant pattern shapes, but their dimensions would still be floating and they would need to be adapted.

And if you look at the number of points of the well-known patterns, it consists of 3 to 6 or so. And they were derived precisely by experience

people. But as there were no computers before, and people could look for patterns even faster, but still only those that are more familiar to the eye.

I think there are many more of these statistically significant patterns, just not as distinctly shaped as the well-known ones.

And shapeless entities are harder to find with the eyes. But in the age of machines, I think we can check that as well.

I remember someone on the spider was already mangling patterns. And he said that the rarer a pattern is, the more likely it is to be executed.

So the rarity of the pattern is an increase in the number of points it consists of and the shape. And because there's more dots, it takes more history

more history to find it, and it's rarer. But its fulfillment is more probable. Impulse cannot grow or decay forever as the market will become predictable after the fact.

That will balance inefficiency/effectiveness.

But by virtue of inertia (or global trends, or whatever) in the market - this same impulse can't always go without damping, which in principle by

logically and should allow you to earn on the principle of following the market after the impulse or series of impulses.

Because of these two examples the market constantly balances between these states and tends to the equilibrium.

to random. But at the same time it can be non-random at some moments. Now, because of all this, when you unbalance

by impulse after stasis and vice versa by stasis after impulse and striving to get back to equilibrium, the imbalance itself can't be

indefinitely. And hence the pattern length of 3-6-10 bars is quite enough to get a statistical advantage. Although it is floating.

That is, with increasing number of points of the pattern, its percentage of working out increases, and it means we can increase the risk level of betting,

or for lovers of Martin-increase bet by bigger coefficient after loss, because

owing to already probable working out of a pattern after losing a bet, the probability of working out the next one increases.

But at the same time the rarity of this pattern increases as the number of points increases on the history, and it turns out that deals are fatter, but they are rarer, which in the end negates the

the advantage of higher probability of working this pattern.

Respectively, the shorter the pattern is, the less fat it is, because the probability of its working is lower - but the more often it will appear on the same

piece of history.

I think there is some compromise between the length of the pattern and its "profitability", "statistical significance", or beatability,

or whatever you want to call it. And this optimal value is just in numbers not big, so working out a pattern with 30 points length, for example,

is not only resource-intensive, but also does not make much sense.

This opens one more interesting thing - by setting the pattern parameters as the target function, you can

set the level of necessary profitability, or vice versa by setting reasonable profitability we may vary the pattern parameters necessary for it

Or maybe he meant the wrong idea what Henium was talking about on one of the forums. Only it is built through the size of the rate.

--------------------------------------------------

However we must understand that a pattern is not a graphical figure as such, it is a set of conditions. Which may or may not have anything

with the form of these conditions, or rather the form to which these conditions will lead. So when I talk above about comparing

the form of the pattern at different levels of fractality, possibly it makes sense, and possibly this comparison should be carried out in a wider sense, when the setting of conditions

of the pattern is not connected with its form, and we will get as a result the same conditions, will give absolutely different forms of the pattern at different

levels of fractality. So, fractality will be shown not in formal form but in some other way (I expounded it non-scientifically but I think the sense is clear).

For example, a row may be fractal by form, size and the number of its components. It may be fractal in size-consisting of 5 bars at each

It may be fractal in size consisting of 5 bars on each level and in this case it may not have the same form. Or it may be fractal in the flatness of the figure, but have blurred outlines

in amplitudes and in "temporal" boundaries. So we can't compare only the shape of the pattern. Or maybe the form is not so important at all

important and "fractality by volatility" is more important when defining the pattern conditions. It must be checked.

===================================================================================================

PS.

But the essence of the branch is not in the wave analysis, we have absolutely different thoughts, it just so happened that these thoughts are located in a branch with such topics. And you have to read it from the 4th page.