The article is interested in how newsmakers exploit numeric records in order to anticipate the future. As this nascent area of data journalism is experimenting with predictive analytics, we examine its reports and computer-generated presentations, often infographics and data visualizations, and ask what time frames and topics are covered by these diagrammatic displays. We too interrogate the strategies that are employed in order to modulate the uncertainty involved in calculating for more than one possible outlook. Based on a comprehensive sample of projects, our analysis shows how data journalism seeks accuracy but has to cope with plenty of prospective probabilities and the puzzle of how to address this multiplicity of futures. Despite their predictive ambition, these forecasts are inherently past-bound because they rest on archival data. This form of quantified premediation limits, we conclude, the range of imaginable ways of future-thought to one preferred mode, that is, extrapolation. Read the full open access article here