The simplest way is to record just one strum, of course. But I guess that will sound quite robotic when repeated over and over again. A better approach (a little bit more complicated) would be to record multiple strums for that slight human variations and use slices with an random lfo targeting the slice number afterwards to play them back.
The OT can automatically slice a sample into 2, 3, 4, 6, 8, 12, 16, 24, 32, 48 or 64 slices. So it’s best to use one of these numbers for how many strums you will record (its just easier and way faster to use the “create slice grid” function of the audio editor than doing all the slicing manually).