Response spectra for California earthquakes of 3.0≤M<7.5 are well described by a simple stochastic ground‐motion model using an equivalent point‐source concept. We determine the best‐fit stress parameter of each California earthquake in the Next Generation Attenuation‐West 2 database based on matching simulated to observed response spectral shapes over a wide frequency range, and we derive an expression for the mean stress as a function of magnitude and focal depth. A calibration factor is calculated for each event; this constant is the required amplitude adjustment in order that the simulations match the observed response spectral amplitudes with zero bias. The best‐fit simulation model suggests that the attenuation in California can be modeled as R−1.3 at distances <50 km and R−0.5 at further distances; this does a better job at matching attenuation trends than the traditional model 1/R model at distances <50 km, particularly for M<5.5 events. The model requires an overall multiplicative calibration factor of Csim=3.16 in order for the simulations to match the observed response spectral amplitudes, for all magnitudes and distances. The calibration constant could be attributed to simplifications inherent in the modeling of source, attenuation, and site processes, and the lack of consideration of multiple phases in stochastic simulations. We conclude that the equivalent point‐source simulation method with the proposed modeling parameters can predict average ground motions in California, generally within a ±25% error band, for magnitudes up to M 7.5, distances <400 km, and frequencies >0.2 Hz. We finish the paper by providing a recipe for developing a simulation‐based generic ground‐motion prediction equation that can be adjusted for source and attenuation attributes in different regions by simple modifications to its key source and attenuation modeling parameters.