Next generation space telescopes, such as the Nancy Grace Roman Space Telescope, HabEx, and LUVOIR would have dramatically increased capabilities for directly imaging exoplanets. A significant fraction of observing time for these mission concepts is likely to be dedicated to orbit determination, with a major goal of constraining the semi-major axis and, consequently, the amount of stellar flux received by the planet. These measurements provide valuable information concerning the formation, evolution, and habitability of directly imaged exoplanets. Here, we demonstrate a novel technique for optimizing the effectiveness of orbit determination follow-up visits in order to maximize the orbital constraints for directly imaged exoplanets. Utilizing Markov Chain Monte Carlo statistical fitting techniques, we fit a suite of orbits to sparse planet position data using the REBOUND N-body integrator. We calculate the time and location of maximum orbital uncertainty, and, subsequently, determine a range of times over which revisit observations will be most effective at constraining the orbit of the planet. Our method is effective for a wide range of planet parameters and is scalable to multi-planet systems. Critically, our technique is significantly more effective at constraining planet orbital parameters than the often-used approach of spacing revisit observations.