I would like to share the link - page 3

New comment
 
LeoV:
This suggests that there are some patterns. Not always and not everywhere - and that is understandable. Which can be used in trading accordingly.
The question is how. I have not found any meaningful statistical relationships in the vicinity of emissions, but maybe I didn't look hard enough)))
 

alsu: Вопрос в том, как.


Deck this is the most important question in trading ))))
 

Here are some statistics. Maybe someone will have some thoughts.

So, take eurusd_h1 for the year, 6736 observations . Here is the chart:

I take a window of 118 bars (that's one week) and start calculating the stats below. Shifted by one bar, counted, etc. I got the graphs. On the mountain axis they are shifted to the beginning, as it is not clear to which point among 118 the statistics of this section belong.

So:

Standard deviation:

Skewness (asymmetry).

Kurtosis. The value =3 corresponds to the normal law.

Jarque-Bera index. If equal to zero, the distribution is normal.

The probability that the distribution is normal:

Probability of non-stationarity of the original quotient (extended Dickey_Fuller unit root test)

Probability of non-stationarity of the first quotient difference (extended Dickey_Fuller unit root test)

The surprising result is that the 1st difference is stationary! Could it be wrong?

 
faa1947: So:

Standard deviation:

Skewness (asymmetry)

Curtosis. Value =3 corresponds to the normal law

The Jarque-Bera index. If equal to zero, the distribution is normal.

The probability that the distribution is normal:

Probability of non-stationarity of the original quotient (extended Dickey_Fuller unit root test)

Probability of non-stationarity of the first quotient difference (extended Dickey_Fuller unit root test)

The surprising result is that the 1st difference is stationary! Could it be wrong?


How are you going to trade with these "cocks"?

What are you going to find or prove?

That there are regularities? So it's pretty clear.

How are you going to use all these graphs to find patterns?

 
LeoV:


How are you going to trade with these "bullshits"?

What are you going to find or prove?

That there are patterns? So it's pretty clear.

How are you going to use all these charts to find patterns?

Actually, these are my questions to the team. I wrote above that I saw references to methods of predicting from the statistical characteristics of the sample. I drew it, so what. If you increase the sample size, all the graphs will be smoother. But I don't see anything but more evidence of non-stationarity in the graphs.
 
faa1947: Actually these are my questions to the collective.
If only someone knew the right answers to the questions )))))
 
LeoV:
If only someone knew the right answers to the questions posed )))))
As always, hope for a freebie. Someone at my link will read the relevant articles and chew on them. That's what this thread is for. Full of links about tails. People are writing for some reason. But it's not clear.
 

Interesting article by U-MIDAS: MIDAS regressions with unrestricted lag polynomials

Link

Many traders use three Elder windows. Here we consider a similar approach. There is a senior timeframe, e.g. quarterly. The new quarterly data is reported once a quarter and with a lag. Question: can I get quarterly data by its monthly (daily) values?

 

faa1947: Зачем-то люди пишут. Но не понятно.

People are always writing something ))) But it's not always clear why.... apparently they just know how to write....))))
 
LeoV:
People are always writing something ))) But it's not always clear why.... apparently they just know how to write....))))
Can't understand it because they're too lazy to figure it out
12345678
New comment
Reason: