The luminosity gap between the two brightest members of galaxy groups and clusters is thought to offer a strong test for the models of galaxy formation and evolution. This study focuses on the statistics of the luminosity gap in galaxy groups, in particular fossil groups, e.g. large luminosity gap, in an analogy with the same in a cosmological simulation. We use spectroscopic legacy data of seventh data release (DR7) of SDSS, to extract a volume limited sample of galaxy groups utilizing modified friends-of-friends (mFoF) algorithm. Attention is paid to galaxy groups with the brightest group galaxy (BGG) more luminous than \Mr = -22. An initial sample of 620 groups in which 109 optical fossil groups, where the luminosity gap exceeds 2 magnitude, were identified. We compare the statistics of the luminosity gap in galaxy groups at low mass range from the SDSS with the same in the Millennium simulations where galaxies are modeled semi-analytically. We show that the BGGs residing in galaxy groups with large luminosity gap, i.e. fossil groups, are on average brighter and live in lower mass halos with respect to their counter parts in non-fossil systems. Although low mass galaxy groups are thought to have recently formed, we show that in galaxy groups with 15 galaxies brighter than $M_r\ge -19.5$, evolutionary process are most likely to be responsible for the large luminosity gap. We also examine a new probe of finding fossil group. In addition we extend the recently introduced observational probe based on the luminosity gap, the butterfly diagram, to galaxy groups and study the probe as a function of halo mass. This probe can, in conjunction with the luminosity function, help to fine tune the semi-analytic models of galaxies employed in the cosmological simulations.