Discussion of article "Random Forests Predict Trends" - page 3
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
At the moment it is profitable ?
The example in the article is demonstrative, extremely redundant.
If we talk about random forests in particular and the application of machine learning models in general (neural networks belong to this class of algorithms), they are widely used in trading.
PS. There are a lot of NS supporters on the forum. So these are not the most efficient algorithms for trading. Random forests are much more efficient.
The example in the article is of a demonstration nature, extremely redundant.
If we talk about random forests in particular and the application of machine learning models in general (neural networks belong to this class of algorithms), they are widely used in trading.
PS. There are a lot of NS supporters on the forum. So these are not the most efficient algorithms for trading. Random forests are much more effective.
Where have you been?
Haven't heard from you in a while?
Where have you been?
Haven't heard from you in a while?
Do you have any links to more in-depth material?
Do you have links to more detailed material?
There is a lot of literature on this topic, mostly in English.
I have written a book "Predicting Trends", in which classification problems are discussed in much more detail than in the article. The book contains literal translations of technical documentation (about 30%), examples of using classification models on the Forex market (about 20% of the text) and explanation of the ideology of building classification models. It also describes the sequence of steps to build an Expert Advisor based on classification models.
More details in the attache.
PS. The book has a rather extensive list of literature on the subject.
vlad19492014.11.23 15:
Dear Vlad!
All arguments about the efficiency of machine learning algorithms make sense under one condition: the model is not retrained. In my practice, it is very difficult to get an untrained model. in particular in this paper, it is an overtrained model.
The overtraining of the model arises because of an erroneous set of predictors, the whole dog is buried in them. Therefore, all efforts should be directed to the selection of predictors and then to the model.
It seems to me that I have managed to find formal signs of predictors' suitability for a particular target variable. If you are interested in this, I will be glad to discuss it in private.
vlad19492014.11.23 15:
Dear Vlad!
All arguments about the efficiency of machine learning algorithms make sense under one condition: the model is not retrained. In my practice, it is very difficult to get an untrained model. in particular in this paper, it is an overtrained model.
The overtraining of the model arises because of an erroneous set of predictors, the whole dog is buried in them. Therefore, all efforts should be directed to the selection of predictors and then to the model.
It seems to me that I have managed to find formal signs of predictors' suitability for a particular target variable. If you are interested in this, I would be happy to discuss it in private.
тттттттттттттттттттттттттттттттттттттттттттттттттттттттттттттттттттттттттттт
Of course I'm interested. Write
vlad19492014.11.23 15:
http://a ppliedpredictivemodeling.com/blog/2014/11/11/some-thoughts-on-do-we-need-hundreds-of-classifiers-to-solve-real-world-classification-problems
I posted this link with annotation on my blog. No one was interested in it. Here is the article itself
Overtraining of the model occurs because of the wrong set of predictors, the whole dog is buried in them. Therefore, all efforts should be directed to the selection of predictors and then to the model.
The model and the selection of predictors are interrelated. First, one should select a model, and then select predictors based on this model by screening by the same model those predictors that have the least "usefulness" in prediction. Although many articles and textbooks teach otherwise: first we select predictors using some method of calculating the relationship between these predictors and the target series - the output. The most common methods of screening are correlation coefficient between predictors and output and mutual information. Then a model is selected usually unrelated to how the predictors were selected. If you think about it (and econometrics textbooks won't tell you this, you have to think for yourself), the method of selecting predictors by correlation coefficient with output essentially selects those predictors that will have the smallest error in a linear regression model (LRC). The method of selecting predictors by their mutual information with output essentially selects those predictors that will give the lowest error in the Nadaraya-Watson regression based model (abstruse name GRNN).