Transforming the data such that one can build targets and predictive features to fit the model.
To build a machine learning model for a trading strategy we need to transform the data from ticks and bars into features and targets. The following sections describe how targets can be constructed using the triple barrier labeling method or the trend scanning method. The other sections describe which type of features the user can generate. The user can construct more traditional technical analysis features such as moving averages, but it is also possible to construct more advanced features that analyse the microstructure of the market.