Searches for periodic variable stars are susceptible to seasonal aliases caused by sampling. Nightly observations from ground based telescopes produce a large number of false detections at the integer multiples of day−1 frequency. Here we discuss the case of VISTA Variables in the Vía Láctea-VIrac VAriable Classification Ensemble (VVV-VIVACE) ID 533558, classified by the VIVACE using machine learning as an EA/EB binary with a period of P = 0.6659 day, in contradiction with a previous classification as an AB-type RR Lyrae with period P = 0.4992 day by the OGLE survey. We discuss the problem of phase coverage causing the misclassification with a wrong period by the VVV, in spite of a robust set of 652 data points over six years of observations. As automatic light-curve classifications of large data sets become more popular, this is a problem that cannot be overlooked.