Point #3 is important. Time-shifting significance estimation is powerful in this context. As the event rate is expected to be very low, almost all the data acq

Point #3 is important. Time-shifting significance estimation is powerful in this context. As the event rate is expected to be very low, almost all the data acquired are noise. Any true gravitational-wave signal must match within ~10 milliseconds of lag between LIGO Hanford and LIGO Livingston. One can shift the clocks of one detector with respect to the other over months of data to look for a matching non-coincident detection. The rate at which those occur allows the straightforward determination of a data-driven false-alarm rate, independent of the analysis technique used. Figure 4 is the relevant one: https://journals.aps.org/prl/pdf/10.1103/PhysRevLett.116.061102 https://journals.aps.org/prl/pdf/10.1103/PhysRevLett.116.061...