Learning curves can bring important insight during the design process of a machine learning model.

We have been researching the selection of variables enterings. When there are too many variables, the model has a worse ability to generalize and is more prone to errors.

We have recently implemented the method Feature Binning and Quantile Transformation. Due to upgraded data preparation, our ML models now achieve better results.

We bring more detailed results, including graphs and comparisons with a benchmark, in our case the S&P 500 Index.

Our goal is to keep only variables that are sufficiently different from each other. Therefore, we removed the mutually correlated series.

SFA software, which analyzes financial articles via NLP, newly shows sentiment statistics.