Context. Certain lines in spectra of the Galactic microquasar SS 433, in particular the brilliant Balmer Hα line, have been interpreted as emission from a circumbinary disk. In this interpretation the orbital speed of the glowing material is in excess of 200 km s −1 and the mass of the binary system in excess of 40 M . A very simple model for excitation of disk material is in remarkable agreement with the observations, yet it seems that the very existence of a circumbinary disk is regarded as controversial. Aims. We investigate whether analysis of optical data from Hα and He I spectral lines in terms of a model, in which the disk is excited by radiation from the close environment of the compact object, can further illuminate the origin of these split spectral lines. Methods. A model in which the excitation of any given patch of putative circumbinary disk material is proportional to the inverse square of its instantaneous distance from the compact object was constructed and compared with published spectra, taken almost nightly over two orbital periods of the binary system. The Hα and He I lines were analysed as superpositions of Gaussian components.Results. The new model provides an excellent description of the observations. The variations of the Hα and He I spectra with orbital phase are described quantitatively, provided the radius of the orbit of the emitting ring is not much greater than the radius of the closest stable circumbinary orbit. The observed variations with orbital phase are not consistent with an origin in a radially expanding ring. Conclusions. The new analysis has greatly strengthened the case for a circumbinary disk orbiting the SS 433 system with a speed of over 200 km s −1 and presents supposed alternative explanations with major difficulties. If the circumbinary disk scenario is essentially correct, the mass of the binary system must exceed 40 M and the compact object must be a rather massive stellar black hole. This possibility should be taken seriously.