About patches, I've seen that too big a size (not enough to cover the bullet anyway) can change a "milli-bit" the accuracy but especially the velocity: the wedge effect excess of the patch works as a brake and the velocity at the muzzle is lower and more irregular than with a good size, so especially at a relatively long distance, something like 100 yards, the scatter cone could be bigger (regardless of lubrication). Of course, this isn't really noticeable at 50 yards, let alone 25 (can't say because at my range, it is forbidden to shoot a rifle at a short distance like 25 yards or meters).
You can see this with the chronometer, but at the muzzle when the patch is open, there is no great difference in the accuracy.
I only do this with .45 caliber, simply because all my rifles are in that caliber. For example, in a Pedersoli Tryon caliber .45 and 1:47 twist, my best compromise is: 1.1" diameter and .010" thickness (0.25mm), for a Dikar (one of the old names of the Tradition brand), for an equal caliber and a twist rate 1:66", the thickness is .013" (0.34mm) and the ball .440" : the two barrels are different...
The diameter value of the patch can create some variation in the "behavior" of the bullet, but it is not so important, on the other hand the adapted lubrication and these are important: barrel (bit rusty or mirror), torsion, type and quality of grooves ...
I think that in the present case the variation is particularly due to the change of position of the shooter, even on the mental side: there are moments with and moments without...