It would describe the observed behavior. Especially if there were an internal quant/efficiency team that wasn't as diligent about regressions as the primary model team.