This chapter describes the basic mechanics for building a forecasting model that uses as input sentiment indicators derived from textual data. In addition, as we focus our target of predictions on financial time series, we present a set of stylized empirical facts describing the statistical properties of lexicon-based sentiment indicators extracted from news on financial markets. Examples of these modeling methods and statistical hypothesis tests are provided on real data. The general goal is to provide guidelines for financial practitioners for the proper construction and interpretation of their own time-dependent numerical information representing public perception toward companies, stocks’ prices, and financial markets in general.