A statistical method is reported to determine the pairs of fragment ions in a mass spectrum that are most susceptible to control by adaptive optimization of the laser pulse shapes in the strong-field regime. The proposed method is based on covariance analysis of the mass spectral fragmentation patterns generated by a set of randomly shaped pulses. The pairs of fragment ions that have higher negative covariances possess a correspondingly higher degree of controllability in an adaptive control experiment, whereas the pairs that have higher positive covariances possess correspondingly lower controllability.