KIDSpec, the Kinetic Inductance Detector Spectrometer, is a proposed optical to near IR Microwave Kinetic Inductance Detector (MKID) spectrograph. MKIDs are superconducting photon counting detectors which are able to resolve the energy of incoming photons and their time of arrival. KIDSpec will use these detectors to separate incoming spectral orders from a grating, thereby not requiring a cross-disperser. In this paper we present a simulation tool for KIDSpec’s potential performance upon construction to optimise a given design. This simulation tool is the KIDSpec Simulator (KSIM), a Python package designed to simulate a variety of KIDSpec and observation parameters. A range of astrophysical objects are simulated: stellar objects, an SDSS observed galaxy, a Seyfert galaxy, and a mock galaxy spectrum from the JAGUAR catalogue. Multiple medium spectral resolution designs for KIDSpec are simulated. The possible impact of MKID energy resolution variance and dead pixels were simulated, with impacts to KIDSpec performance observed using the Reduced Chi-Squared (RCS) value. Using dead pixel percentages from current instruments, the RCS result was found to only increase to 1.21 at worst for one of the designs simulated. SNR comparisons of object simulations between KSIM and X-Shooter’s ETC were also simulated. KIDSpec offers a particular improvement over X-Shooter for short and faint observations. For a Seyfert galaxy (mR = 21) simulation with a 180s exposure, KIDSpec had an average SNR of 4.8, in contrast to 1.5 for X-Shooter. Using KSIM the design of KIDSpec can be optimised to improve the instrument further.