Since the early 2010s, investigators and antitrust regulators have focused on communications among market-making traders in online chatrooms as part of criminal and civil investigations. Traders have often used these forums to post quotes, negotiate trades, and share market information, and competition authorities have cited materials from these chats as evidence to support claims of anticompetitive conduct. But because transcripts of these communications can run into the millions of pages, it is often prohibitively difficult to analyze them and extract the relevant content manually. Recently, however, powerful data science methods – in particular, machine learning (ML) and natural language processing (NLP) algorithms – have emerged; these algorithms can process vast quantities of text and identify complex linguistic patterns in order to obtain pertinent information without sacrificing analytical precision. In this article, four Analysis Group consultants explain these new tools and describe in detail the ways in which they can help identify relevant evidence with greater precision than has previously been the case. Using blinded examples from recent litigation, the authors offer guidance for using these NLP and ML algorithms, as well as practical steps for organizing, structuring, and analyzing large datasets.