An intriguing question in the context of dynamics arises: Could a moon possess a moon itself? Such a configuration does not exist in the Solar System, although this may be possible in theory. Kollmeier et al. (2019) determined the critical size of a satellite necessary to host a long-lived sub-satellite, or submoon. However, the orbital constraints for these submoons to exist are still undetermined. Domingos et al. (2006) indicated that moons are stable out to a fraction of the host planet Hill radius $R_{H,p}$, which in turn depends on the eccentricity of its host's orbit. Motivated by this, we simulate a system of exomoons and submoons for $10^5$ planetary orbits, while considering many initial orbital phases to obtain the critical semimajor axis in terms of $R_{H,p}$ or the hosts satellite's Hill radius $R_{H,sat}$, respectively. We find that, assuming circular coplanar orbits, the stability limit for exomoons is 0.40 $R_{H,p}$ and for a submoon is 0.33 $R_{H,sat}$. Additionally, we discuss the observational feasibility of detecting these sub-satellites through photometric, radial velocity, or direct imaging observations using the Neptunes-sized exomoon candidates Kepler 1625b-I (Teachey et al. 2018) and identify how stability can shape the identification of future candidates.