Roman Poshtar / 个人资料
- 信息
|
10+ 年
经验
|
2
产品
|
4023
演示版
|
|
3
工作
|
0
信号
|
0
订阅者
|
Majors CatBoost Clusters Real EA operates on the CatBoost algorithm using clustering methods. CatBoost is fed a set of features based on price and moving averages (MA) . What exactly: Distances Price - MA (for different periods) MA − MA → show the deviation and strength of the trend Normalization Division by price or MA → makes features scale-independent Window statistics (WINDOWS) Average, std, max, min → reflects volatility and behavior over a period Lags
Majors Perceptron Distance works on a perceptron recruitment algorithm. The selection and construction of features for training are described in detail in my articles, which you can read below: Experiments with neural networks (Part 1): Revisiting geometry https://www.mql5.com/en/articles/11077 Experiments with neural networks (Part 2): Smart neural network optimization https://www.mql5.com/en/articles/11186 Experiments with neural networks (Part 3): Practical application
https://www.mql5.com/ru/market/product/118772

