So, how is regression used in stock trading? In practice, using regression to predict a stock’s return is difficult because the signal and financial data is low compared to the noise. Moreover, the models are sensitive to some choices you make about the model. For instance, you’ll have to decide how much of your previous data to use since more recent data is more relevant than very old data. Regression models are pretty sensitive to these design choices. Regression is also sensitive to outliers in the data, as it adds more noise to the training data. However, there are still important reasons to learn regression to analyze stock returns. Learning regression is useful because we can apply the same regression techniques to analyze time-series data. Time series analysis looks at data that is collected at regular intervals over time and uses that to predict its value in the near future. One more note about why regression is important to learn, if we learn the details of how neural networks work, many of the same principles apply. In fact, if we think of regression as a fundamental building block, then a neural network is a building that is composed of many regression building blocks.