a trading strategy based on Elliott Wave Theory - page 88
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
One more thing. Something I don't understand is this: СКО2/3[N]=({D[N]-D[2N/3]}/{N-2N/3})^0.5
As I see it, in your notations, RMS2/3[N]=(D[2N/3])^0.5
Or, if you try to represent it as a difference:
СКО2/3[N]=({S[N]-S[последняя треть]}/{2N/3})^0.5
The RMS is the root of the variance (sum of squares of deviations divided by some number). Then the sum of variance squares on bar 100 minus the sum of variance squares on bar 33 will yield the sum of variance squares from bar 33 to 100. The rest is easy. In general, you are right, as I understand it, I must have expressed it wrongly.
The point is that two channels can have the same within statistical significance regression error variance, but different price variance, roughly speaking, one channel will be steeper, the other will be flatter. The question is which channel to choose. Bulashev considers three criteria for assessing the quality of a regression line, all of which involve the ratio of the above two variances. Choosing among these three criteria is indeed a personal creativity, and choosing the variance of the regression error to compare the quality of the approximation is not entirely correct.
You further reply to
Vladislav takes the worst of this class.
If you mean that Vladislav takes the widest confidence interval possible for this, then you misunderstand me, if something else, then I misunderstand you.
So according to solandra's algorithm after "...the sample does not fall outside the 99% confidence interval. The channel that has the smallest RMS value is selected from the series of going bar to bar. I was asking - how to select the smallest RMS if they can be statistically indistinguishable or it's all a trifle?
Next
The choice of accuracy of the bracket is determined by the statistical accuracy of its definition. You said yourself that it is a random variable.
Honestly, I don't get it, or you don't get me. I didn't actually ask about choosing the accuracy of the sko. I must have misunderstood. I was wondering why the sko of the sample is compared to the sko of 2\3 sample to determine the convergence, and not the sko of some other part of 5\8, 7\9 etc. Will this have a significant effect on the selection results? Or are these again insignificant details? The borderline between good and evil?:)
If you are interested in a working model, then take all this as an axiom, implement this model programmatically and the market itself will show you whether your set of axioms is fair or not.
I mean, "what's there to think about, you've got to shake it". It seemed to me that the beauty of the approach under discussion is that before looking at what the market will show, a thorough analysis is carried out and the criteria for distinguishing what the market can show in general are substantiated. I'm not very good at programming, and I'm not good at programming without understanding what exactly is needed and then seeing what happened. I must be really picky about small things.
Regards
И еще. Что-то я не понял вот это: СКО2/3[N]=({D[N]-D[2N/3]}/{N-2N/3})^0.5
The RMS is the root of the variance (the sum of the squares of deviation divided by some number). Then the sum of variance squares on bar 100 minus the sum of variance squares on bar 33 will give the sum of variance squares from bar 33 to bar 100. The rest is simple.
That is, you should probably write the formula RMS1/3[N]=({D[N]-D[2N/3]}/{N-2N/3})^0.5 or RMS2/3[N]=({D[N]-D[1N/3]}/{N-1N/3})^0.5.
Did I get it right?
If I understand you correctly, this is not correct as the deviations for 2/3 are counted from another regression line. You try building a channel at a certain length and another one at 2/3 and you will see that the lines don't coincide and hence the sum of the deviations will be different (maybe that's what you meant?). As far as I understood the variance or RMS itself cannot be used to calculate subsequent values since every new bar gives a new line and changes the entire variance, in theory it cannot be calculated from the variance obtained on the previous bar. I seem to have managed to take it into account in this cycle and even the channel plot by two thirds looks ok (when regression coefficients are calculated we also calculate the sum of CB squares and the sum of CB itself therefore we can use them to calculate the dispersion on the next bar, but I failed to use the dispersion itself) but when I made the RMS file and looked more carefully I saw incomprehensible things that pop up on every 3 bars.(although I seem to have taken into account the unequal movement of 2/3 interval bounds)
PS If anyone is interested I can post the RMS data (corrected :))
И еще. Что-то я не понял вот это: СКО2/3[N]=({D[N]-D[2N/3]}/{N-2N/3})^0.5
СКО - это корень из дисперсии(суммы квадратов отклонений, деленной на некое число). Тогда сумма квадратов отклоений на баре 100 минус сумма квадратов отклоений на баре 33 даст сумму квадратов отклонений от 33 до 100-го бара. Дальше все просто.
So the formula should probably be written as RMS1/3[N]=({D[N]-D[2N/3]}/{N-2N/3})^0.5 or RMS2/3[N]=({D[N]-D[1N/3]}/{N-1N/3})^0.5
Both formulas you wrote down here are wrong. The variance and sum of squares are different. D[N]=S[N]/N
Given this, the variance of the difference of the intervals is not equal to the variance of the intervals. Rosh's comment, as I understand it, expresses agreement with my clarification.
The point is that pure mathematics and the market are substantially different things. It seems to me that we should proceed from this. For example:
The analysis is certainly thorough, but it particularly comes from the fact that Vladislav assumes that it is impossible to accurately predict the subsequent price movement. But it is possible to make a "non-random prediction", the probability of which will be somehow (!) equal to its actual probability. Generally speaking, a thorough analysis is an experimentally verified theory. Vladislav's approach is not a theory, but only a model. And its experimental verification is just now underway, the results on the empire. What do you want from those who, like you, have only seen the model on paper and haven't understood everything?
Do not try to "draw out" more from us than we know. We won't admit it anyway :-)
You'd better try to analyze the model on your own and offer us your vision.
The rest is in accordance with what has been said
IMHO. Price dispersions can only differ significantly if the samples differ significantly. If this is the case (i.e. one is significantly longer than the other), then both channels have strength, but differ in their long-runness. These are the kinds of channels you are interested in. Also, I believe it is the error variance that is the main source of information for criterion creation. And discussing the quality of the approximation is, in my opinion, pointless. The ISC gives the best version.
So according to solandra's algorithm after "...the sample does not fall outside the 99% interval. The channel that has the smallest RMS value is selected from the series of going bar to bar. I was asking - how do you select the smallest RMS if they may be statistically indistinguishable or it's all a trifle?
IMHO. Don't assume that Vladislav and solandr think alike. Solandr only shared his understanding. You can take the channel with the smallest sko, or you can take the whole class with the same statistical significance. And use the worst one out of it.
IMHO.2/3 is Vladislav's choice. Don't wait for justification. Try other options. There are scissors present here. The larger the fraction, the more likely you are not to fall out of the channel on the current bar. And the greater will be the delusion you are thereby leading yourself into. And if you take "the less," you will get a tougher condition and increase the probability of falling out of the channel prematurely. That is, your criterion will throw you out of the channel before it actually collapses. You can assume that 2/3 is a parameter to be optimised.
Good luck.
Thank you
Parabola Y(X)=Ax^2+Bx+C , the coefficients of the parabola are in no way related to the linear regression coefficients Ah+B.