The strain-dependent kinetics of the myosin working stroke under load is derived from a flat-energy-landscape model for its untethered lever-arm, and compared with other scenarios in the literature. The "flat landscape" scenario is compatible with muscle-fiber experiments, but is more critically relevant to single-myosin experiments with an optically trapped actin filament. In such experiments, the strain dependence of stroke kinetics may be explored by comparing event-averaged and time-averaged displacements of the filament. With a specific kinetic model of the cross-bridge cycle, we have previously shown that the event-averaged displacement underestimates the working stroke. Here we predict that the two kinds of averaging give diverging estimates of the working stroke as the resolving time of the event detector is decreased to 1 ms or less, the discrepancy being critically dependent on the strain dependence of the stroke rate. Such analysis of trap displacement data offers the possibility of testing the strain-dependent stroke rate predicted by the flat-landscape model.