From theory to practice - page 162

 
Alexander Sevastyanov:

Is it possible to explain everything scientifically? Yesterday morning in Davos Trump said that he was in favour of a weak dollar, which benefits American importers and the economy as a whole, which was in principle in line with the weakening of the dollar after his election. The market reacted instantly to his words and the dollar fell 1,000-2,000 5-digit points against other currencies trending with no discernible correction. In the evening Trump explained that his words were misunderstood and the dollar recovered within an hour or two.

This is the scientific explanation - the market movements had a specific, physically existing reason. It is not a formula, but a pseudo-scientific one - it looks abstruse, but it is far from the essence.
 
bas:
This is the scientific explanation - the market movements had a specific, physically existing cause. It's not a formula, it's a pseudo-scientific explanation - it looks abstruse, but it's far from it.
It reminds me of something. Times of T.D.Lysenko and his ilk. The rhetoric is the same.
 
Yuriy Asaulenko:
Something reminds me of this. The times of T.D. Lysenko and his ilk. The rhetoric is the same.
Here, I am absolutely sure that your favourite neural networks will not help in this case. Definitely.
 
Alexander_K2:
Here, I am absolutely sure that your favourite neural networks will not help in this case either. Definitely.

In your case, the NS really can't help. But there's nothing for them to do here.)

Your case is quite simple. Why are the outliers on your trend going out of the distribution boundaries?

 
Yuriy Asaulenko:

In your case, the NS really can't help. But there's nothing for them to do here.)

Your case is quite simple. Why do outliers on the trend go beyond the limits of the distribution?

And infinitely many times... It's an unbelievable picture! After all, on the whole (at a huge general population of data) relative to 0, the motion of the wave packet forms an almost normal distribution. But, this particular plot - I don't understand it yet. I would understand just a giant outlier from 0, but many, many times crossing a large enough confidence level...

Shit, Yuri - don't upset me with my own stupidity. I'm out of here.

 
Alexander_K2:

And infinitely many times... Unbelievable picture! After all, in general (with a huge population of data) relative to 0 the motion of the wave packet forms an almost normal distribution. But, this particular plot - I don't understand it yet. I would understand just a giant outlier from 0, but many, many times crossing a large enough confidence level...

Shit, Yuri - don't upset me with my own stupidity. I'm out of here.

Especially for you, for the umpteenth time I'm posting a picture.

This is the same price chart (normalized from 0 to 50), plus on Z is the probability density. Here you see how distributions move over time.

That is, a distribution is formed, then there is a "jump" in price outside that distribution, and a new distribution is formed around another centre.

The average will naturally pull up to this new centre. Where will it go.)) And because you look everything relative to the average and collect the information you can say "for centuries", you get a good sense of continuity of the process.

In my opinion, this picture answers all your questions.

 
Alexander_K2:

And infinitely many times... Unbelievable picture! After all, in general (with a huge totality of data) relative to 0 the motion of the wave packet forms an almost normal distribution. But, this particular plot - I don't understand it yet. I would understand just a gigantic outlier from 0, but many, many times crossing a large enough confidence level...

Shit, Yuri - don't upset me with my own stupidity. I'm out of here.

Don't feel bad, there's more than a few bison in the forex business who've had their teeth broken. Some of them are gone, and some continue to fight it with desperate determination and fanaticism. And only among them are individuals who, after many years, finally succeed. If you are not satisfied with that path, it is better to give it up).
 
Alexander_K2:

Incredible picture! After all, in general (with a huge population of data) relative to 0 the motion of the wave packet forms a near-normal distribution. But, this particular section - I don't understand it yet. I would understand just a giant outlier from 0, but many, many times crossing a large enough confidence level...

Normal (and any other) distribution doesn't mean that the price getting to its 99% level will come back) I wrote about this back on page 1 of one of your previous threads)

And you're not building it relative to zero at all, your reference level itself shifts relative to price (be it SMA, or channel borders, or whatever). Which is what happens in case of trends, which you stubbornly ignored)

 
Nikolay Demko:

I say we need a feature that signals that the trend is changing from trend to flat and vice versa.


Nothing is signalling this, no characteristics.

 
Alexander_K2:


They are right, you can't do anything without taking the trend into account, and I told you the same thing earlier. My Expert Advisor takes into account both the trend and the deviation from the centre. Although it caused some turbulence today, all trades closed with profit. Here are the results from Monday.