Electrooculogram (EOG) studies in sport have focused on temporal oculomotor features such as the quiet eye (QE) period – the duration of the final fixation on the action’s visual target. On the one hand, it is unclear whether EOG can provide valid QE measurements due to its poor spatial resolution relative to the eye tracker (ET). On the other hand, EOG’s high temporal resolution facilitates extraction of time-frequency content through spectral decomposition. In this study, we aimed to (1) examine which EOG signal processing options and algorithms (eye position or velocity) yield the most valid QE measurements, (2) introduce a novel method—the EOG spectrogram—that describes both temporal and spectral oculomotor features, and (3) rank the utility of four oculomotor measures (QE-ET, QEEOG-pos, QE-EOG-vel, EOG spectrogram) for predicting motor performance on unseen data. We co-recorded EOG and ET while 16 participants of varying expertise putted golf balls to a 4-m distance target on a flat surface. Concurrent validity and temporal discrepancy analyses revealed that QE-EOG-pos and QE-EOG-vel are valid and accurate for certain processing options (channel, filter, thresholds). The EOG spectrogram—obtained through multi-taper Fast-Fourier-Transform— distinguished activity in lower frequencies (saccades) and higher frequencies (fixational activity) before and during movement execution. Nested cross-validation indicated that the EOG spectrogram yielded the lowest generalizability error and the greatest stability on unseen data, followed by a tie between QE-ET and QE-EOG-vel, and then by QE-EOGpos. Correlational analyses suggested a monotonic association between better motor performance and greater EOG high-frequency fixational activity (> 25 Hz) during movement execution.