Gravitational Waves (GW) from stellar-mass BH binaries have recently been observed by LIGO, but GW from their supermassive counterparts have remained elusive. Recent upper limits from Pulsar Timing Arrays (PTA) have excluded significant portions of the predicted parameter space. Most previous studies, however, have assumed that most or all Massive Black Hole (MBH) Binaries merge effectively and quickly. I will present results derived—for the first time—from cosmological, hydrodynamic simulations with self-consistently coevolved populations of MBH. We perform post-processing simulations of the MBH merger process, using realistic galactic environments, including models of dynamical friction, stellar scattering, gas drag from a circumbinary disk, and GW emission—with no assumptions of merger fractions or timescales. We find that despite only the most massive systems merging effectively (and still on gigayear timescales), the GW Background is only just below current detection limits with PTA. Our models suggest that PTA should make detections within the next decade, and will provide information about MBH binary populations, environments, and even eccentricities. I'll also briefly discuss prospects for observations of dual-AGN, and the possible importance of MBH triples in the merger process.