Trend indicator based on artificial intelligence

I offer research system, made in the form of indicator in which to forecast the behavior of a financial instrument used machine learning methods.

This system can predict the direction of the current trend through set by the user, a period of time and to assess the accuracy of the forecast.
Works on any symbols and timeframes, provided sufficient training data (set by user). All the mathematical calculations, including training of prediction models, are produced in terminal R. a Detailed description of the research system and its parameters You can find in "Instructions for use" (attached .pdf file).
This product is the result of many hours of work by programmers, but like any research platform, it requires a lot of time spent on tests and experiments. Joint efforts it can be done faster and better. In the near future we plan to expand the functionality of the system (addition of new prediction algorithms, adding a forecast of the lifetime of the current trend and its volatility).

Classification of machine learning algorithms used in the system:

1. By appointment
· Predicting the direction of the trend the main signal (the probability up or down)
· Predicting the quality of the main signal (the probability is high, uncertainty is low)
· Predicting the direction of the trend based on the ensemble of algorithms (stacking model)

2. Type
· Extreme Gradient Boosting (xgboost)(xgboost.readthedocs.io/en/latest/)
· andom Forest (rf)(ru.wikipedia.org/wiki/Random_forest)
· C5.0 algorithm (C5) (ru.wikipedia.org/wiki/C4.5)
· DeepBoost algorithm (deepboost) (research.google.com/pubs/pub42856.html)
· Multivariate Adaptive Regression Spline (earth) (en.wikipedia.org/wiki/Multivariate_adaptive_regression_splines)