Discussion of article "Random Forests Predict Trends" - page 6
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
No, not ordinary flud, but flud on the topic and in order to increase the quality of articles on this forum. Otherwise, they write chapters from books on modelling various processes not related to the market, and leave their applicability to the market for the reader to check. I declare with authority that all these methods are not applicable to the market. I have already proved it myself. And those who write articles and books without such proofs have one goal: to earn credits. There are two types of participants on the market: those who try to earn by trading and those who try to earn by teaching the first ones how to trade and earn on the market without having any successful experience in trading. The latter annoy me, especially when they authoritatively try to sell you either an unproven model or a proven model, but not suitable for the market, and give homework to the readers to check it themselves. They say that we are teachers, and you students are oaks, go and check our theories.
I would like to draw your attention that I not only write articles and books, but also offer services to bring your set of predictors to a real EA. Attach again. I am posting here the list from the atach.
As a practical implementation of the book, I offer the following services:
1. Consultations on forming a source file for Rattle:
Free of charge
2. Conclusion on the overfitting (over fitting) of the proposed set of predictors:
3. Conclusion with justification of the availability of predictors on which it is possible to build the model without overfitting (if any):
4. Code on R model without overfitting:
5. Participation in writing an Expert Advisor using the R model:
Traders who do not have the necessary training in EA programming and econometric models can trade with the help of software robots on financial markets (forex, stock exchanges) by implementing the following plan:
1. Independent selection of predictors
2. With my help, testing predictors and building models with a guarantee of no overtraining
3. Ordering an Expert Advisor that uses trading signals from the model created by me.
You are wrong, gpwr.
In my opinion, the purpose of any article is to arouse interest in a topic, subject, field of knowledge.... Most readers want to get a ready-made, universal solution here and now, with little or no effort. faa1947 writes about prediction and classification methods, which is very rare on this resource....
If you argue, then argue in a reasoned manner. For example, presenting your model or its parameters...
It is unconstructive to poke at the fact that "I checked, it didn't work, show me your successful version"....
I personally don't need a ready-made solution. Tell me about the essence of the model, even without theoretical details and step-by-step guide, and I will implement it myself. But I will have zero motivation if the author does not show at least some practical results of this model, such as the percentage of classification accuracy on untrained history. I assure you that in the vast majority of articles published here, the accuracy will be 50%. No respectable journal in the world will publish a paper without practical tests of its theory.
And look what you offer me: present your model and prove that it is better. How about asking the author to show the results of his model? All these articles are written in Yusuf's style: describe the theory, publish the article without results, then look for programmers, then find out that the theory doesn't work, then add martin, grid and other additives and get "soup from an axe". Although from the point of view of attracting the public these articles do their job. The more controversial the theory, the more heated the discussion. The site achieves its goal by the number of guests, which in my opinion is a more important goal than presenting something applicable to trading.
I agree that there is not much constructive in my posts. But agree that this article was written for the purpose of advertising. There is not much constructive in it. So do not expect constructive discussion.
How about asking the author to show the results of his model?
Holy crap!
For you personally, copypaste from section 7 of my article:
We get the following results:
Holy crap!
For you personally, copypaste from section 7 of my article:
We obtain the following results:
By results we mean not these results, but the results of trading the TS based on the model.
This is not an econometrics forum.
by results we do not mean these results, but the results of trading the TS based on the model.
This is not an econometrics forum.
That's what your client wanted:
But I will have zero motivation if the author does not show at least some practical results of this model, such as the percentage of classification accuracy on untrained history.
Your defendant, unlike you, has the requisite knowledge, on which he quite rightly asked for"classification accuracy percentage" and not some profit factor.
Yes, there is a gap in the article, it is not finished, which is why I offer my services. I offer to all, but only to those who have knowledge of the topic.
PS.
And the article is from what field of expertise? That is the one we are discussing.
PS .
Are there any real TCs without econometrics?
That's what your client wanted:
Except that I will have zero motivation unless the author shows some practical results of this model such as the percentage of classification accuracy on the untrained history
Your defendant, unlike you, has the requisite knowledge, on which he quite rightly asked for"classification accuracy percentage", not some profit factor.
Yes, there is a gap in the article, it is not finished, which is why I offer my services. I offer to all, but only to those who have knowledge of the topic.
PS.
And the article is from what field of expertise? That is the one we are discussing.
PS .
And that there are real TCs without econometrics?
Sorry. I was wrong, I didn't see the results. I take all my words back. Good luck using this model in trading!
I'm sorry. I was wrong, I didn't see the results. I take all my words back. Good luck using this model in trading!
On the basis of the discussed article, a collective has started to form outside the forum, which is trying to fight against overtraining (overfitting) of Expert Advisors. In the paradigm of TA + tester this problem cannot even be identified, and therefore the depo drain is just a matter of time. And if the depo is not drained, it is just luck, like, it happened that way, with no guarantees for the future.
With Rattle it is quite easy to identify the overtraining of the model, which is determined by a set of predictors, not by the model. The choice and application of a particular model is a tenth matter. The cost of learning Rattle itself is ridiculous, but Rattle allows you to see the problem of overtraining and then maybe solve it for this particular set of predictors.
Are there any real TCs without econometrics?
None of the real TCs I've seen had anything to do with econometrics.
Is it weak to just make money on your models and not sell your opinion on predictors?
On the basis of the discussed article, a collective has started to form outside the forum, which is trying to fight against overtraining (overfitting) of advisors. In the paradigm of TA + tester this problem cannot even be identified, and that's why the depo drain is just a matter of time. And if the depo is not drained, it is just luck, like, it happened that way, with no guarantees for the future.
And here is the level of the conversation. Retraining is not a problem at all. I mean, EVERYTHING. it's easy to identify and easy to solve, without econometrics.
But if there is no baby in the bathtub to begin with, nothing will help.Here comes the connoisseur, we haven't seen you for a long time... Go dump your bitcoins (if you haven't dumped them all yet), instead of talking about econometrics....