From theory to practice - page 1543
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
Well, here's your data:
Yes, you have a stationary process on the sum of the increments, but alas, it's utter nonsense.
You have a price series with one increment and yours is completely irrelevant. That's completely out of line... I might as well take any BP and write next to it the increments from the GCF.
Price, or indeed any process, is integral to the increments and nothing else. They are two inseparable things.
strange ... I keep noticing that my robot is opening orders on some invisible line (the red line).
maybe that's your true channel median)
I don't know :)
Do you do any BP transformations or do you work with the original and just observe the statistics?
I don't really do everything the usual way, it's kind of related to statistics but the detection methods themselves don't work according to statistical formulas at all.
like this one line split into two lines again
and then back into one...
hmm... strange...
maybe you can explain it based on your channels...
I don't like it when I don't understand something...
of course I could be wrong...
human eyesight is such an analyzer))
I don't know :)
Do you do any BP transformations or do you work with the original and just observe the stats?
no, i only work with pure original price and i do not transform it in any way.
I only analyse its movements.
That's why I don't have any mismatch errors at all.
I used to work closely with statistics...but eventually I realised that absolutely any transformation leads to misalignment.
Any transformation at all.
that's how the market works...
and if you convert the price you're chasing your own tail...
you improve the formulas, the disparity goes up and down, but it never goes away... unfortunately.
and it's the random wandering that's to blame.
Your lowest level of self-education prevents you from engaging in discussion. That's all I can say.
....
If one manages to reduce the real series of increments to the normal one using Lambert's W-function, so that the set of cumulative sums also forms a normal distribution, i.e. a stationary random sequence according to Kolmogorov, then such series can be predicted, no matter what one says.
....
Are we talking about this function?
https://dic.academic.ru/dic.nsf/ruwiki/1172968
And why is it a panacea? Or is it just another shot at it?
zy
In my opinion, the formulation of the problem"to reduce the real series of increments to the normal, so that the set of cumulative sums also forms a normal distribution" is wrong, or rather, in such a formulation the problem has no solution.
Sash, take a look at this file, put it in your system. But you have to do an increment to convert it.
The question is this.
You have the most beautiful pictures of probability densities in the market. And you are trying, by studying them, to create your own market theory. As a matter of fact, I do the same. But is it good? After all such densities are found only in nuclear physics, describing uncontrolled processes of nuclear reactions.
Is it not better to reduce this economy to known distributions - Gauss type?
The Box-Cox transform (God invented surnames! Ugh.) does not work in the market, everyone knows that. No one has studied or applied the Lambert W-function transformation.
Maybe you know something else? Or is there no point in any transformations?
I used to get into such thickets...hypercomplex planes of event probability...cointegrating segments...scale-time transformations...wavelets... nothing works....any work on market event prediction ends up with "blah blah blah"...and that's it. there are no complete works on really working mechanisms.
that's why i don't particularly like mathematics although i use it privately.
we could of course use the Monty Hall paradox...if only we could bring the binary event to three consecutive outcomes rather than two. And that's impossible...unfortunately.Yeah.
Honestly - I only heard about it here for the first time.
It is the hypothesis that when we get a transformed stationary series, we get to investigate it with a set of cumulative sums. I've already told you how to do this, but in fact without transformations this strategy is pouring into the market. Alas, non-stationarity is an extremely serious thing and it's hard to fight it.
If you suggest any other methods for obtaining a stationary BP from the real one, I would be grateful.
There are no such methods. And not because they have not been found. But because it is fundamentally impossible. Figuratively speaking, stationarity are small islands in a huge boundless ocean of nonstationarity.
Yeah.
Honestly - I only heard about it here for the first time.
It is the hypothesis that when we get a transformed stationary series, we get to investigate it with a set of cumulative sums. I've already told you how to do this, but in fact without transformations this strategy is pouring into the market. Alas, non-stationarity is an extremely serious thing and it's hard to fight it.
If you suggest any other methods for obtaining a stationary BP from the real one, I would be grateful.
Some applications of the Lambert function
http://trudymai.ru/upload/iblock/98a/primenenie-funktsii-lamberta-v-teorii-turbulentnogo-treniya.pdf?lang=ru&issue=50
http://edu.secna.ru/media/f/LambertW.pdf
I mean pure BP and its stats?
My opinion is that this is such a well-trodden topic that it is incredibly difficult to make a profit on it. We need some kind of breakthrough.
Lambert's function seems to me to be such a breakthrough. Unfortunately, it doesn't exist in VisSim... Unless you write it yourself...
I'd like just a peek at the converted series....
Like on the files - here's the original unsteady BP, and here's the transformed one with Lambert. And I'll show you how to profit.
don't forget that you're converting absolutely the entire time series, and that's a fundamental error as I've been saying for a long time.
I think I have a grail, but not because it brings 100% profitable trades, but because, when you change its parameters, it adapts not to the history of price movements
i can catch fatter tails, lesser ones, or random fluctuations.
Fitting to statistics is fitting strictly to a probability distribution.
it means that it affects the quality of all trades at once, both on history and in the future.
that's the main part of the grail.
That's exactly what you're trying to do. You're confusing fitting to the history of quotes and fitting tothe probability distribution.
That's what you're trying to do. You're only confusing the fitting with the history of the price movement and the probability distribution.
It is natural because the truth is very simple - history does not repeat itself unlike statistics.
it's very simple.