The Laser Interferometric Gravitational-wave Observatory (LIGO) has detected gravitational waves from the merger of two 30Msun black holes (Abbott et al. 2016). An earlier prediction (Belczynski et al. 2010a) that binary black holes would constitute the first detection was therefore borne out, but predictions of merger rates have varied over several orders of magnitude (Abadie et al. 2010). We have generated and analyze a suite of models, each with 640 million simulations of black hole formation and mergers under a range of environmental conditions and physical processes. We find that, given the sensitivity of current gravitational wave detectors, the most common detections will originate from two comparable-mass coalescing black holes, with total masses in the range 20-80 Msun. Within the framework of our simulations, we conclude that the stellar progenitors of the LIGO event GW150914 were formed in an environment where the metallicity was less than 10 per cent of solar, and had initial masses of 40-100 Msun. They interacted through mass transfer and a common envelope phase, and both black holes formed without supernova explosions. The black holes' spins and orbital angular momenta are nearly parallel and nearly unchanged since birth. The progenitors likely formed either at 2 Gyr, or somewhat less likely at 11 Gyr after Big Bang. The field formation of binary black holes, with low BH natal kicks and restricted common envelope evolution, produces 40 times more binary black holes than dynamical formation channels involving globular clusters.