Gravitational waves from isolated sources have eluded detection so far. The upper limit of long-lasting continuous gravitational wave emission can now probe physically motivated models with the most optimistic being strongly constrained. Naturally, one might want to relax the assumption of the gravitational wave being quasi-infinite in duration, leading to the idea of transient continuous gravitational waves. In this paper, we outline how to get transient continuous waves from magnetars (or strongly magnetized neutron stars) that exhibit glitches and/or antiglitches and apply the model to magnetar SGR 1935+2154. The toy model hypothesizes that at a glitch or antiglitch, mass is ejected from the magnetar but becomes trapped on its outward journey through the magnetosphere. Depending on the height of the trapped ejecta and the magnetic inclination angle, we are able to reproduce both glitches and antiglitches from simple angular momentum arguments. The trapped ejecta causes the magnetar to precess leading to gravitational wave emission at once and twice the magnetar’s spin frequency, for a duration equal to however long the ejecta is trapped for. We find that the gravitational waves are more detectable when the magnetar is: closer, rotating faster, or has larger glitches/antiglitches. The detectability also improves when the ejecta height and magnetic inclination angle have values near their critical values, though this requires more mass to be ejected to remain consistent with the observed glitch/antiglitch. We find it unlikely that gravitational waves will be detected from SGR 1935+2154 when using the trapped ejecta model.