Advisor based on fundamental analysis - page 3

New comment
 
Yes, such an Advisor will take at least 2 years to make =), about half a year just to search, then half a year to test these FAs =) so I think this is an unrealistic idea
 
wenay:
Yes, this EA will take at least 2 years to build =), about half a year to find it, then half a year to test these FAs =) so I think this is an unrealistic idea.

Give me your reasons. Why half a year?
 
I think we need to dig in the direction of multivariate statistical analysis. Any other ideas?
 
The idea is interesting, but several problems can arise and you have to think of a way of solving them in advance.
The first one that came to mind:
  • problems of identifying the news. Even if we highlight a figure, it is difficult for us to estimate it automatically. For example, the quid's refinancing rate cut - was by 50 basis points and the euro shot up, recently by 75 and the euro flew down.
  • testing problems. The test results show that the best time to enter the market is when the news releases, but in real trading or even in demo trading the slippage may be as large as 20-30 pips.
  • Problems of determining the importance of news or indices. For example, 25 news have been released per day. How to determine the ones that are more important, that did not influence anything and that slowed the movement.
Personally I have not managed to answer these questions yet and therefore I use FA only as an additional flag for calculating stops and takeoffs. The Expert Advisor does not solve FA problems, I do.
 
Plus additionally try to lay down some basic principles of classical FA in the form of some rules.
 

The speed of this case is mainly determined by the speed at which the text file is created. Then there is the indicator - the moment of its publication, there are values: current, previous, and all other previous values. For the beginning, there is an indicator that shows publication points of a certain indicator and its value. Then the Expert Advisor and optimization will show whether there is something in it or not. Anything can be done, if only there were such a file with indicators.

 
Kharin:
It's an interesting idea, but there could be a few problems, and you have to figure out how to solve them beforehand.
The first thing that comes to mind is:
  • problems of news identification. Even if we highlight a figure, it will be difficult for us to assess it automatically. For example, a decrease in the refinancing rate of the quid - was by 50 basis points and the euro shot up, recently by 75 and the euro flew down.
  • Problems with testing. According to the test results, the optimal entry will be at the moment of news release, but on the real and even on the demo it is possible to have serious slippage of about 20-30 pips.
  • Problems of determining the importance of the news or the index. For example, 25 news have been released per day. How to determine which ones are more important, which ones have had no effect, and which ones have slowed down the movement.
Personally, I have not been able to answer these questions yet and therefore I use FA only as an additional flag for calculation of stops and takeoffs. The Expert Advisor does not solve FA problems, I do.


1) I need some kind of news reference book, which can be updated and changed. It will contain a list of news, their identifiers, names, degree of importance by default (because the news importance can change by idea). Since news identifiers or names can change for different sources, we need to make a separate table of correspondence between the identifier in the directory and the identifier or key in the source.

2) Not sure if we should enter the market during or based on the news release. The FA will give us a general trend for some period. Or you can wait for some pause (set filter by time of news release and wait for some time until everything settles down). By the way, the optimal moment of entering the market can be found at the stage of testing and analysis of historical data. Although, the slippage cannot be estimated then. But we can take into account the statistics on time and slippage value.

3) The importance of the news is indicated in many sources. For example on http://www.dailyfx.com/calendar. And I think after analyzing the historical data you can estimate the degree of importance of the main news and assign them some weights.

 
If people are interested and willing, we can cooperate and make such an indicator/advisor through joint efforts and ideas.
 
I also can't imagine (by the way, does anyone?) an algorithm for processing a number that characterises a news item. Suppose we already have a database with news identifiers already set. Then we can define a new index value, compare it with the previous one and ... then what? Then we need to evaluate the comparison result. The question arises: "How? Eventually, the Expert Advisor should contain lines like

if(...) OrderSend(OP_BUY...)
else if(...) OrderSend(OP_SELL...);
else return(0);

It's just that it happens from time to time, and the same news can be received in different ways at different times. As it happened with the Eurobucks, for example. With all that, the analysts predicted that the rate will go down if the rate stays above the rate. And they were right. But how to make the machine make the same conclusions? After all, with the same change, it had to open in opposite directions.
 
kailex:

I have an idea to develop an EA (or a separate indicator) that displays general market sentiment according to fundamental analysis indicators...

Your comments and suggestions are welcome.

Mr. kailex, why do you need it?

Or do you think it is possible to write a profitable EA based on these indices and FAs?

12345678910...12
New comment
Reason: