The ringdown of a perturbed black hole consists of a superposition of quasi-normal modes (QNMs), with complex frequencies determined by the black hole's mass and spin, while phases and amplitudes depend on binary parameters. Traditional semi-agnostic ringdown analyses ignore the remnant's binary-merger origin, assuming instead generic perturbations. This can lead to an unphysical number of free parameters and usage of priors inconsistent with those imposed by full inspiral-merger-ringdown (IMR) analysis. Here we revisit an alternative approach, first proposed in Calderón Bustillo et al. 2021, in which ringdowns are modeled using the post-merger portion of full IMR waveform models. This naturally includes all QNMs without adding extra degrees of freedom and ties priors to binary parameters. We analyze the signal GW150914 with the post-merger portion of the IMR surrogate model NRSur7dq4, finding decisive evidence for the no-hair theorem with a Bayes factor >650: 1 or >99. 8% probability, compared to only 3: 1 from semi-agnostic spectroscopy, consistent with previous work. We also find modest evidence (5: 1) for the (, m, n) = (2, 2, 1) overtone and none for higher modes. Using simulated signals, we validate our formalism by showing accurate recovery of remnant properties. Next, by analysing simulated signals with post-merger signal-to-noise ratios up to 100, we show that classical spectroscopy based on overtone detection alone cannot provide strong no-hair tests. Finally, we demonstrate that the non-orthogonality of QNMs makes the inferred mode content strongly model-dependent, leading to multiple inconsistent interpretations that are equally well-supported by the data. We briefly discuss potential implications for recent events, such as GW231123 and GW250114.
Chandra et al. (Mon,) studied this question.