Numerical relativistic codes are now able follow the orbits of black holes from inspiral to merger, finally. Back in the 60′s, Peres suggested that asymmetries in the configuration black hole binaries (different masses or spins) would have the net effect of beaming gravitational wave radiation in some preferred direction, a problem theoretically analog to electromagnetic radiation recoil. In the gravitational case the recoil arises from the interference of the mass quadrupole and the mass octopole (or alternatively, the flow quadrupole radiation.) The magnitude of the kick velocity depends on: a) the mass ratio of the binary, b) the spin magnitudes and c) spin orientations. The Baker et al. 2008 numerical fit looks like this:
where mu = q/(q+1) and q =m1/m2 < 1, A,B,H,K, and Phi_i are constants. The maximum possible recoil of velocity V ~ 4,000 km/s can be achieved when the kick is imparted in the direction parallel to the angular momentum vector for maximally spinning, equal mass black holes with counter-aligned spins oriented in the direction parallel to the orbital plane (Campanelli et al. 2007). Less rare recoil velocity distributions are given by Baker et al. 2008 Table 3 and in Figure 2 of Tanaka and Haiman 2009 (a very nice paper, sans possible typo in equation 12.) The above configuration can be represented schematically by the figure below.
There are several aspects of the recoil process that are still unclear, such as its effect on the cosmological build up of massive black holes from seeds to the SMBHs we see at the centers of galaxies today. The holes are launched into orbits that depend on the details of the host potential:
a) In spherical bulges the black hole will oscillate right through the center, and return to the center after a few dynamical times, or deplete / “heat up” the stars in the nucleus, reducing the efficiency of dynamical and extending the return time.
b) If the black hole is able to leave the bulge and enters a triaxial dark matter halo, it may not return at all within a Hubble time (we are about to publish a paper that looks at this in detail.)
c) High resolution simulations of gas mergers (e.g. Mayer et al. 2007) show that up to 60% of the initial gas is funneled to the center of the remnant. In that particular simulation, the mass of the nuclear disk is Md = 3e9 Msun, and recoling black holes with velocities of ~500 km/s are not able to escape the central region (scale length L=75 pc).
Because the majority of black holes merge at high redshifts, detection becomes difficult. The most commonly used technique to find off-set AGN / QSOs is to look for shifts between the broad-line region associated with the QSO and the narrow-line region associated with the starlight from the galaxy. Unfortunately no recoil candidate has been confirmed (one must be brave to report a recoiling SMBH these days, since their paper will be heavily cited by people suggesting alternative hypotheses.)
If LISA….
When two black holes merger, any asymmetries in their initial masses and / or spins can cause gravitational wave radiation to be beamed in a preferred direction. To conserve linear momentum, the coalescing black hole binary suffers a recoil “kick” that can range from 100-4000 km/s. The idea that this phenomenon can eject massive black holes (MBH) from their host galaxies has been around for quite some time now, but now offset nuclei have actually been confirmed.
3. Download Gadget2 from Volker Springel’s website 
