This is a project in progress meant to identify wide, astrometric binaries in Gaia. These are binaries with orbital periods too long for Gaia's binary's working group to identify them based on astrometric orbital motion. Instead, we are attempting to identify these binaries by matching pairs of stars in phase space - using position, proper motion, and distance. The method relies on a Bayesian formalism, in which a calculation of the evidence is included. Such a calculation is necessary to fully differentiate between true binaries and randomly aligned, optical doubles. Furthermore, this method should excel at identifying the widest separated pairs - potentially a degree in separation. We train on previous proper motion catalogs such as the revised NLTT catalog.