Gathering a team to develop an IO (decision tree/forest) in relation to trend strategies - page 7
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
To make a decision - signal to input/filter signal. A pattern is looked for.
This is logical, but as long as I've searched independently for patterns, it has always turned out that we have patterns on the history, but WHEN these patterns will appear in the future is not determined in the future, i.e.
- we have found patterns on the history, after such a pattern, we should always open an order in a certain direction and when will this pattern appear in the future? - on certain parts of the history it can be repeated 10 times, then it will be absent for years....
- they work in a certain part of the history only in the +, and then they work like any other indicators - then you guess, then you don't....
if you were looking for WHEN the patterns work or WHEN they stop working, it would probably be something new, but here.... whether it's a forest or an indicator, imho the labor intensity is different, but the output is the same
This is logical, but as long as I have searched independently for patterns, it has always turned out that we have patterns on the history, but WHEN these patterns will appear in the future is not determined in the future, i.e.
- we have found patterns on the history, after such a pattern, we should always open an order in a certain direction and when will this pattern appear in the future? - on certain parts of the history it can be repeated 10 times, then it will be absent for years....
- they work in a certain part of the history only in the +, and then they work like any other indicators - then you guess, then you don't....
if you were looking for WHEN the patterns work or WHEN they stop working, it would probably be something new, but here.... whether it is a forest or an indicator, imho the labor intensity is different, but the output will be the same
You are right to say so, but for some reason you put regularities in different groups when they make up a whole. The fact of presence of that about what you speak, certainly exists. but here the matter is either that the regularity was false or that the dependence of occurrence of this regularity is not revealed, i.e. this regularity is more complex, than it seemed at first sight. Personally I think that it is possible to struggle with such effects only by taking longer period on which we investigate the pattern, then either this pattern will disappear on it, or remain, but if it remains, then we can expect that it will remain for indefinite time - until more complex structure of this pattern will not change. There is also such a method as cross-validation, this is if we catch a large piece of history on which the pattern operates and do not take into account other factors affecting this pattern, i.e. when the pattern is simplified, then using cross-validation (dividing the sample into different parts, including in a random way) will help you assess the stability of the pattern on different intervals and if this pattern is not stable and there is no way to make it more stable, then such a pattern will be rejected.
The advantage of the tree is to solve the problem for each part of the sample separately, i.e. dividing the sample into sections, whereas applying indicators is continuous to the whole testing range (sometimes it is not, but in simple codes it is), this approach allows to sift good inputs from bad ones, but this sifting turns out to be more stationary than applying the decision tree.
Since no one wants to work for the idea, I'm looking for co-investors who are willing to contribute to the tasks described.
A community of anonymous loggers...
You haven't even decided on the tools. On a small historical range, the quality of the trees planted is questionable. And to grow a jungle, the power of a single computer is not enough. Without Spark, I don't believe in this venture.
Closed groups can now be created here to communicate in private. I propose to move there or to slack and add only normal guys, and then the flooders will get it. Now on vacation, in the 1 st numbers will be digging a little again, already have a promising system. I.e. if anything I'm in, at least for the exchange of experience
Glad you are interested in collaborative activities.
How do you organise these groups?
However, as long as there are hypothetically only two people it is difficult to talk about a group, i.e. we should wait to go into a closed space for now, maybe someone else would like to work in this direction.
A community of anonymous loggers...
You haven't even decided on the tools. On a small historical range, the quality of the trees planted is questionable. And to grow a jungle, the power of a single computer is not enough. Without Spark, I don't believe in this venture.
Again, I don't know what tools we are talking about. What do you mean by this word, software, algorithm, programming language?
About power - I have small power - 3 Phenom II x6, 2 FX-8350, 1 Phenom II 5x - total power to start, there is a hypothetical rig, which can be transferred from mining to working with trees, but I need a software - I'm looking for people who can do it. Now I'm thinking of the best way to combine all the power into one cluster. And each hypothetical participant has at least one computer...
;))) closed groups...
What can I say, everyone thinks in terms of his or her own depravity. Better join in.
Again, I do not understand what tools we are talking about. What do you mean by this word, software, algorithm, programming language?
All of it.
Now it's more like "Guys, I heard that scaffolding and networks are cool! Let's try them!"
First, study the tools themselves.