a trading strategy based on Elliott Wave Theory - page 51
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
The idea is certainly an interesting one, it has crossed my mind too. However, there is one subtlety here. If I understand you correctly, NE is an average value. If this is the case (and even if it is not), I would like to know how you can use history to estimate channel stability? The difficulty I see here is this. As you rightly pointed out, there are two parameters on which the lifetime of a channel (most likely) depends - slope angle and width. If they didn't exist, you could just create a statistical series of all channels on the history and calculate the average and sko for it. And having a sko - estimate the probability that the lifetime of a given channel is out. :-) Then we could (like we do now with Murray levels) draw vertical lines whose intersection with the channel lines would give additional information about the reversal zones for each confidence interval. However, there are these two values - slope angle and width - which makes it impossible for us to compare the lifetime of the two channels if they have different values. I think a solution to this problem does exist, but a CORRECT SETTING OF THE PROBLEM is needed. As a person far from mathematical statistics I address to experts: dear
Vladislav and others, can you take the trouble to formulate the formulation of this problem?
NE is a random variable. Of course for each dip angle and width of the channel we may simply find experimentally this distribution (by sampling). But this is not very objective and may lose statistical significance, or discretization will be too crude. The second option is analytical and for it the already considered Hurst coefficient is the most suitable. Indeed, it takes into account both distribution statistics and sample size (actually an analogue of time). That is, we can consider the value of Hurst coefficient for a channel. If it is close to 0.5, the channel is not statistically confirmed, but if it is too high for a channel, there is a high probability that it is already "overripe" and will soon fall apart. I.e. the whole task comes down to analysis of the pair: level of Murray + coefficient of Hurst of the channel that crosses this level. For this pair, we may gather statistics like: probability of penetration of the level 4/8 by a channel with the Hurst 0.75 (to be analyzed with necessary accuracy, for example, with 0.05) = 0.8, etc. Then the combinations found must be checked for stability. Some of them will be non-stationary and it does not make sense to use them, although the theoretical probabilities of penetration or rebound may be high for them. Testing for a breakout or rebound is quite simple - the main criterion is keeping or leaving the channel. In other words, what is stronger the channel or the level.
I.e. the Hurst coefficient is a general and exhaustive measure of the channel evaluation and has properties we need: the narrower the channel and the bigger the slope angle, the faster this coefficient will grow for this channel with increasing time of price being in it.
There has been a big discussion about Hurst here. If you are referring to the standard procedure for calculating it, then your suggestion can only give something if it is time-dependent. However, as far as I understand Hearst, it should not change its value in the channel.
But even if I am wrong, the need to formulate the task correctly still remains. The only parameters in your suggestion will be Hurst and Murray levels. Although I personally like the angle and the cant as parameters better.
There has been a big discussion about Hurst here. If you are referring to the standard procedure for calculating it, then your suggestion can only give something if it is time dependent. However, as far as I understand Hearst, it should not change its value in the channel.
But even if I am wrong, the need to formulate the task correctly still remains. The only parameters in your suggestion will be Hurst and Murray levels. Although I personally like the angle and the cant as parameters better.
Yes, I guess you are right. You will not get away with Hurst alone here :)
The mistake was that by this condition the channels were discarded instead of being used.
That picture is an illustration of what happens if the channels are badly sampled :o)
Angle, as applied to a plot, is not a very convenient notion. If it is expressed in degrees, it is related to the scale in both coordinates. But if it is expressed in terms of pips/time, then there is something to it. But exactly how can it be used in this form?
100 pips per day is only 0.0694... pips per minute, is such a channel steep or gentle?
There is a chapter in Bulashev's book about estimation of forecast lifetime by a single linear regression channel. It remains to generalize this estimation to several simultaneously active channels and connect it with the Murray levels.
But this estimation does not depend on the slope angle in any way.
Angle, as applied to a graph, is a perfectly normal concept. However, the attempt to express it in degrees is acceptable only when x and y have the same dimensionality and hence the calculation of trigonometric functions is justified. In this case the dimensionality is pips/bar. Therefore the angle can only be measured by the LR coefficient.
You are right, however, in that this is not very convenient. As a result of the fact that this coefficient is dimensional its value will change when moving from one t/f to another. And this is not good. :-)
Are you referring to Bulashev's chapter on "Regression analysis"?
By Bulashev, yes, I meant this chapter, more precisely: "8.12 Forecasting based on a single-factor linear regression."
Yes, there is such a thing there - "Forecasting Horizon". But it's not quite the same thing.
The horizon shows how far back the forecast can be drawn from the current bar.
And the lifetime is the absolute length of the trend, independent of the current bar.
Yes, there is such a thing as the Forecasting Horizon. But it's still not quite the same thing.
The horizon shows how far a forecast can be made from the current bar.
And the lifetime is the absolute length of the trend, independent of the current bar.
I believe the number of bars the price spent in the channel is its internal time. And the exhaustive characteristic of a channel is still the Hurst coefficient. It contains the slope angle as well as the channel width (implicitly through sigma, spread and N). I.e. we can consider a threefold: Hurst coefficient, Murray level, N-number of bars inside the channel. In other words, we should consider channels with the same persistence level as identical ones.
1) The level of importance of channel selection criteria is the same (i.e. you find an optimal combination of these criteria), or there is a sequential selection from more important to less important criteria.
2) Having read a lot of clever books, I forgot what to find :). If I have understood correctly that you mean under a concept of potential energy functional, it is not clear why we search for it as the result of search will be the equation (not value, but function!) of a trajectory at movement along which change of potential energy (during movement, instead of at achievement of an end point!I understand that the price moves along this very trajectory and we have already chosen the equation that approximates this trajectory (regression equation), it remains only to conclude how well we approximate this trajectory. But if we look for it anyway, we may find a quadratic function and if the coefficients В and С in equation Ах^2+Вх+С are equal (or very close) to those in the regression equation, it may be the necessary channel, but I have already gone into doubt :)
I used to think I was good at maths, now I see that it's not so :(
By the way, did you notice that Alex Niroba has disappeared again. He promised to show something cool and not again, and it's a pity... :)
So, if we have a criterion to estimate the trend strength, we can rather reliably suppose after which wave a channel breakout will come. I think the reversal zones by Vladislav as well as support/resistance levels by Murray, for example, are also very powerful tools for such estimation. It is clear that when a breakthrough has occurred a new channel should start from the top of the last wave. In my opinion, this is quite an algorithmic approach.
Hearst's Criterion is exactly the criterion of a trend's strength. When several small channels (waves) form one large channel (large wave) - you can probably measure in it. My script and the indicator do not allow to build several channels automatically yet. However, breaking of the channel boundary (in the context of this branch) usually leads to reversal (in my memory) or flat (I usually build channels with the script once a day and see the results 7-10 hours later). At that moment the optimal channel suddenly becomes very wide, it can be used as well. You cannot build from the last vertex due to limitations of the minimum sample or you may need to go lower in the frame.