Machine learning in trading: theory, models, practice and algo-trading - page 2091
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
There are a lot of problems here.
For example, what do you want to find?
formalize, what should happen? (or not to happen, so to say the null hypothesis .... I picked up the terms in class))) )
and in general, such task, imho, faster to look in database, like to unload ZigZag? with dates in one table, and in the second parsing news with categories of importance
But then again... what do we want to take as the null hypothesis?
My approach is standard) the null hypothesis is price behaves like SB. If we are talking about a zigzag, the value Z=z/m-1, where z is the length of the zigzag knee and m is its parameter, in the case of SB has an exponential distribution with a single parameter. This will be the null hypothesis. Accordingly, the alternative hypothesis will be a rejection of this statement.
There are a lot of problems here.
For example, what do you want to find?
formalize, what should happen? (or not to happen, so to say the null hypothesis .... I picked up the terms in class))) )
and in general, such task, imho, faster to look in database, like to unload ZigZag? with dates in one table, and in the second parsing news with categories of importance
But then again... what do we want to take as the null hypothesis?
All in all, the conversation is just about the difficulty of formalizing to simple tasks. And news has different strength and focus, and the effect of news on price has a different response. There are many variants, and there are many stages. You can take the same news and only co-directional news and watch the response, or you can take only multi-directional news and watch the strength of the action. And how to watch, you can look for a change in channel rate, you can zigzag, you can some kind of average, a time scale problem as well. And this all needs to be broken down into simple study patterns.
Well, in general, like a good wine, or cheese to make. It depends not only on the order of what you put in and how much, but also on a lot of other factors that are not visible to the eye and can't be felt).
In my approach everything is standard) the null hypothesis - the price behaves like SB. If we are talking about a zigzag, the value Z=z/m-1, where z is the length of the zigzag knee and m is its parameter, in the case of SB has an exponential distribution with a single parameter. This will be the null hypothesis. Accordingly, the alternative hypothesis will be a rejection of this statement.
In the case of SB has an echr distribution of the fractional length of the zigzag by the number of events per unit time minus 1, i.e., are the events regular per unit time? What is the parameter m is not fully understood.
To fully understand, we are on different languages (levels) ))))
In the case of SB has an eChr distribution of the length of the zigzag by the number of events per unit time minus 1, i.e. the events are regular per unit time? What is the parameter m is not fully understood.
To fully understand, we are still in different languages (levels) ))))
This is just a primitive zigzag - a new knee starts when the price moves more than m from the top of the previous knee in the opposite direction.
Time is there only as the first coordinate of the zigzag knee top - to be able to find the knee corresponding to the event.
This is just the most primitive zigzag - a new knee starts when the price passes a distance greater than m from the top of the previous knee in the opposite direction to it.
Time is present there only as the first coordinate of a vertex of a zigzag knee - to be able to determine, which knee corresponds to the event we are interested in.
now I see).
This is just the most primitive zigzag - a new knee starts when the price passes a distance greater than m from the top of the previous knee in the opposite direction to it.
Time is present there only as the first coordinate of the top of a zigzag knee - to be able to find the knee that corresponds to the event.
OK
So far, only a WP has been formalized - or rather, we have decided that we need a WP with one setting - "points".
about the use of the database - if properly download the data, it is not difficult to make the selection of interest, but we have not yet formalized what we're looking for
I think that you can unload the ZZ in the format:
- bar number, time of the last vertex, length of the current WP beam (i.e. in the future from the vertex) , direction of the beam (up | dn)
Well, the news can be unloaded in any format, the time will be there, but we'll look at the bar number according to the query data from the database
but what should happen due to the news release?
what is the minimal setting of the ZigZag that should determine that the ZigZag has reacted to the news?
what is the TF?
....
imho, there are a lot of technical questions so far to try to assess whether news affects price movement? - What should happen to prove that the price is a SB?
OK
So far, only the ROM has been formalized.
about the use of the database - if you properly unload the data, it is not difficult to make the sample of interest, but we have not yet formalized what we are looking for
I think that we can unload RZ in the following format:
- bar number, time of the last vertex, length of the current WP beam (i.e. in the future from the vertex) , direction of the beam (up | dn)
Well, the news can be unloaded in any format, the time will be there, but we'll look at the bar number according to the query data from the database
but what should happen due to the news release?
what is the minimal setting of the ZigZag that should determine that the ZigZag has reacted to the news?
what is the TF?
....
imho, there are a lot of technical questions so far to try to assess whether news affects price movement? - There will be a spike in valutility, but what has to happen to prove that the price is SB?
Purely one zigzag can be compared to previous zigzags. The length has become longer than the previous average or shorter, but that's rough. TF 15 minutes or an hour, depending on what kind of feedback we are looking for. I like the channels of small zigzags better. But this is a major complication.
Purely one zigzag can be compared to the previous ones. The length has become longer than the previous average or shorter, but this is rough. TF 15 minutes or an hour, depending on what kind of feedback we are looking for. I like the channels of small zigzags better. But this is a big complication.
On the news this "length" can get longer
But we're looking for trends, right? or reversals?
what do we do if there is a price reversal? or it was a hairpin on the news?
Why TF15? The smallest available TF is M1, there is the logic of using D1, but M15 - well, you like the number 15, but it should not be a scientific basis for the process under study?
gentlemen experts... the minute is out... will you take more? ))
yes, give me two!
For the main subject, the news - I want to predict the outcome:
- the effect of the news on further price movement, all the same need to be marked manually
- ZigZag is not suitable for this evaluation, it's better to use some trend indicator... Then why analyze the news? We work the old way, tester + GA or tester + MO
- the task of news analysis is a heuristic task, so to speak, this is creativity