a trading strategy based on Elliott Wave Theory - page 35

New comment
 
As for the symbolic part, this is not exactly true! Read the whole thread carefully! ....... to the method of calculation of the Hurst parameter, which is described in all the books!

Mr Solandr and Mr Vladislav the branch has been read (I speak about me) and some moments have been read several times, the problem is not in reading the branch but in understanding the mathematical (statistical) language, unfortunately I do not speak a higher mathematics, I am silent about the Bulashev's contents (to understand it like that...(no comments)... but do not suggest to graduate another university specializing in NM and stuff like that and that I understand everything :) So I'll risk to ask again (on behalf of those who have not yet understood what all the hocus-pocus), and with your permission I'll allocate moments which (again I speak of myself) is not competent to handle, plz correct me where necessary, if time for it is not sorry.[/quote]

Vladislav: quite a number of linear channels can be plotted at any given time, but only the linear regression channel will have minimal error. For forecasting it is better to take the best approximation at a given time.
...
applicability of the mat statistics apparatus on samples roughly longer than 30 degrees of freedom (bars in this case, but not on any sample !!!!!! - The criterion itself cuts off samples - hence the algorithm turns out to be iterative) - errors there will tend to zero - hence analysis of small periods by such methods is doomed - I think so. When I calculate daily levels on intraday charts the error is small - the sample length is enough.
...
In this case a fractal (i.e. self-similar structure) is a regression channel, of course with an estimated target

Approaching the current price to one of the limits of the linear regression channel (if we continue it at this point in time), but the LR channel should be calculated with at least 30 bars if we build LR channel on Daily, 180 on H4, 720 on H1, etc.

Vladislav: As for the statistical significance of the levels - it will become clear to you once you build confidence intervals - for example, the more oversold, the more likely a return to the equilibrium point and possibly even to the opposite boundary (this becomes clear when approaching the equilibrium point) - so the confidence intervals and cut off the probability levels.

from ch.5 Bulashev, where it says about the calculation of confidence intervals I do not ...I don't understand
I understand: if I conditionally divide an LR channel which is as close as possible to something (from my right or wrong understanding of approximation), then the Confidence Interval is the point where the current price is in relation to the channel width in % ratio or in other words "for example if I take the bottom of an LR channel as 0 and the top as 1, the price is somewhere md 0.01<price<1"
Explain to me if something is wrong.

By contrast, with H > 0.5 events today will matter tomorrow, i.e. the information received continues to be taken into account by the market some time later. This is not simply an autocorrelation, where the influence of the information quickly decreases, but it is a long-term memory, which causes the influence of the information to persist for a long period of time. Of course such influence does diminish over time, but it is still slower than short-term correlations. This influence is characterised by the length of the cycle, when it drops to an indistinguishable value.

You say that information also attenuates both up and down sampling, except with different intensities, so it remains unclear which channel has priority at maximum approximation - the one with the smallest or the largest sample?

Vladislav: So about principles - construction of projection (actually it is extrapolation, which in turn is based on current approximation) just substantiates the order of approximating functions, and this in turn is the way to determine errors. From considerations of field potentiality we infer the order of approximation and the basic laws that our approximation must satisfy and also estimate the maximum permissible error at which our approximation can still be considered adequate.

Error - does it mean finding the price (or something else) within the tolerance level for approaching the LR channel, please explain and how do you measure this error

Vladislav: When you start your research you will face a situation of some uncertainty ........ here you need a quality criterion, on the basis of which you can choose the best approximation in terms of extrapolation from constructed approximations...... I have chosen as such a criterion the minimum of the potential energy functional (and this is also a consequence of the potentiality of the price field). How to build approximations - you have already figured out that it is a quadratic form - you can look for a parabola, or you can do it another way - think about it.
Solandr: Channels are found that satisfy criteria of RMS convergence and non-falling of sample beyond 99% of interval. The channel that has a lower RMS value is selected from the series of channels going bar by bar. Channels are drawn both on the basis of a linear regression and a kvadratic function (a function of the form y=a*x^2+b*x+c, which is decomposed into two functions - the linear regression equation and the parabola, which allows us to estimate errors)

I understand that the trajectory of price movement (like a trend!) is a function expressed in terms of price and time y=a*x^2+b*x+c (tchk), but I do not understand what and where to substitute, where is the price and where is time? but I think that the game is the price :))), but I also need to decompose it, maybe you can show me a clearer way than functionals and parabolas, at least on bricks :)

Vladislav: Next - once you have chosen approximation, you can continue this trajectory into the future (the further you go, the less reliable the result will be).


It's clear that the farther is the closer to the stars, but it's still unclear, what is the limit of forecasting in your system: 1. just the PRICE relative to the current bar to the right; 2. the upper/lower boundary of the channel LEE at the current bar or to the right relative to the current bar

Vladislav: What I do - I choose period of approximation (not the whole sample, but about 2/3, the last third extrapolate and compare with obtained real prices, if it does not fall out of confidence interval, then I use this approximation for further extrapolations, but this is related to implementation and methods of increasing stability of iterative algorithms).

From this I only understand the part where I need to take 2/3, somehow there to compare with the real prices, and the rest I am a dummy. If it is not difficult, translate from mate language to Soviet native language, please

Solandr: Given the existing channels at any point of the price you can calculate the probability of trend continuation/reversal. If I'm doing this, I can't find anything better than to make such an averaged probability calculation for all channels using weights... So I take probability sum for each channel with weights equal to number of bars of each channel divided by total number of bars in all channels and thus find an averaged total probability.

If it means the current price, then why at any point when the current price has one, if the price is from the LEFT side, then why is it there, maybe it was there? Explain it this way (again with bricks): A is this and that, B is another this and that, C is the third and that, D is the fourth and so on, and all together is this and that ABCD :))) Imagine that you are explaining to a Chinese in Russian how a zeppelin flies

Yurixx: I don't know how to do anything at all without knowing what I'm doing.

i strongly agree and repeat the same

Solandr: I trade on real account, but with mere penny lots so far. I'm trading on demo accounts, but I've understood for a long time that it's better to trade one kopeck each on the real account than millions on demo.

i woof the first depo in a ton, even the minimum lot, i mean i do not regret that i did not burn it, i just bought the experience

to sum it up.
1. I have learned a lot for myself and I have been using Murray's indices since they appear on the forum, of course I have my questions but I will discuss them later.
2. Solandr, I also admire you for finally getting to the bottom (or just about) of it and having the nerve and time to explain things and answer questions, thank you.
3. The question of the true calculation of the Hurst coefficient remains open.
4. Also, kudos to everyone who participates in this thread, broken record
 
Теперь я вижу, что ANG3110 собаку съел на таких вещах (ветку на пауке помню) :)

Indeed, I fully support it!!!
Very nice and a Necessary indicator!
ANG3110, could you explain the code? It's hard to get it right off the bat. Maybe you have your own blog on the forum, where everything is written about this indicator in detail? Thanks in advance for the information.

I also want to know how to use it and what Value1 says.
I will also be grateful to you.
 
ANG3110, could you explain the code? It's hard to figure it out at first glance. Maybe you have your own branch on the forum, where everything is described in detail about this indicator? Thank you in advance for the information.

To be honest, I have trouble remembering quickly how I created the code. I wrote it 1.5 years ago using mgl2, and then I just copied it to mql4. So, don't mind me. Of course, if an urgent practical need arises, I'd remember everything.
There is some stuff on the "spider", but it's crude, tawdry.
I have my mail at the beginning of the code and if something interesting you can write.

Regards - Alexander.
 
Теперь я вижу, что ANG3110 собаку съел на таких вещах (ветку на пауке помню) :)

Indeed, I fully support it!!!
A very beautiful and Necessary indicator!
ANG3110, could you explain the code? It's hard to get it right off the bat. Maybe you have your own blog on the forum, where everything is written about this indicator in detail? Thanks in advance for the information.


Here ANG3110 has placed his indicators, you can see that he is fond of numerical methods - http://forex.kbpauk.ru/showflat.php/Cat/0/Number/100566/an/0/page/0#100566

And the code from the previous page, exactly, solves the problem of finding the coefficients of parabola (at m=2) by Gaussian least squares method for an initial bar equal to NULL and for 24 hours (the number of bars in the calculation is 24*60/Period()). (hours=24).
In other words, it is enough to change these input variables and you can use this function.
 
And here's a piece of the parabola from that indicator

 
Made a toy in Excel, very sobering.



than the currencies.



H value less than 0.5 for normally distributed increments could not be obtained.
 
Hi Rosh!
Interesting toy :-) What is it ?
I mean explain what is in columns B,C,D, what data did you use and what is in the graphs.
As far as I understand, each of the two Hearst index values refers to the respective graph as a whole ?
What do
refer to?
<br/ translate="no"> Average 0
Std.deviation 1

on the green field in each of the tables ? After all, judging from the graphs, the average of these data cannot be 0.

Thank you
 
Privet,

Vsio vygliadit krasivo kokda rabotajete s dannymi v istoriji, no naskolko budet pod etimi ras4iotami xoroshy prognozy na budus4ije ceny? :)

Kstate, kokda ja na4al programirovat', immeno parabolu iskal v grafikax, but with boleje primitivnym sposobom - 5 moving averages on periods Xn= Xn-1 * 2

Patom vsio taki pereshol pod EWA+Wolfe Waves i sdelal sebe obs4ije vyvodov iz etix teorii (kak govoritsia, jesli xot' 2 teoriji pokazyvajet na tu ze storonu, mozet i pravda:) )
 
Kstate, nas4iot vsiakix kanalov, takoje u menia toze polu4ajetsia avtomatom, tak kak Waves teorija predskazyvajet to4ki, v katoryjyx risujetsia liniji trenda i targeta,
iz nix obrazujetsia kanal. Tolko raznica mezdu etoj teoriji i vyvodami u menia takoje:

1) Volny s4itajetsia tak kak s4itajetsia volny Elliota, but sameraja nomeracija volny - kak v v vogue Waves from 1 to 6
2) Trend s4itajetsia jesli imejetsia Elliot Wave 3 (in any pattern)
3) Target trendline risethia mezdu na4alom Elliot Wave 2 i 4
4) Time at Arrival sets between Elliot Wave 1 and 4
5) "Sweep Zone" ostajotsia kak v Wolfe Waves teorii - mezdu Target i Price at arrival (v etix to4kax kon4ajetsia trend)

Vot primer kakio eto vygliadit v praktika:


Order mozno vstavliat' na 3,4,5 i 6 (riskovannyje na 2,4,6) pod etim :)
 
Hi, Rosh!
Interesting toy :-) What is it ?
I mean explain what is in columns B,C,D, what data did you use and what is in the graphs.
As far as I understand, each of the two Hearst index values relate to the respective graph as a whole ?
What do the values refer to

Mean 0
Stands Deviation 1

on the green field in each of the tables ? After all, judging from the graphs, the average of these data cannot be 0.

Thank you


I am attaching an Excel file which generates 1000 increments following a normal distribution with expectation 0 and standard deviation 1 (standard class distribution). If you press F9 - each time a new graph will be generated for which Hirst criterion will be calculated.
https://c.mql5.com/mql4/forum/2006/06/HurstExcel.zip

HH. The formulas show the algorithm for calculating this criterion (which I am 99.9999% sure of).
By the way, it is useful to sit and dumbly press F9, it helps cure "market vision" childish illnesses :) The market is not a very liquid instrument, but you can't prove it at once.
1...282930313233343536373839404142...309
New comment