Deal or No Deal: Predicting Mergers and Acquisitions at Scale

While research on merger and acquisition (M&A) has been extensive in the finance literature, in the realm of data science, little work has been done on deploying a successful Big Data informed M&A prediction model. In this paper, we explore what can be learned about M&A activity from a firm's annual Form 10-K SEC tiling. We utilize natural language processing (NLP) techniques to vectorize each filing's textual data. Next, we cluster firms by industry and identify keywords suggestive of upcoming M&A activity. We then train a classifier to predict acquirers and targets, which we use to forecast the most likely M&As of 2019. Lastly, we deploy an application which enables users to query our forecasts and visualize our data.