LOL the 14Z HRRR now has the discrete sups firing in LA and AR, congealing into a QLCS by the time they cross the river. A couple of the overnight runs had an intense discrete supercell tracking from N MS into S TN (almost like a 12/23/15-type track) but that is long gone. The HRRR has really been all over the place with this one.
On a couple of past occasions I've seen it be almost spookily accurate with the location and timing of a tornadic supercell almost from the first run that was in range, but this is not going to be one of those events.
I think it has to do with the fact that we're dealing with (the potential of) open warm-sector supercell development, which is triggered by very subtle features the timing and location of which is very difficult to pin down, where as the ones it really nailed were storms forced by easily trackable synoptic features like a triple point/warm front. Also, maybe something was tweaked in the model after 5/20/19 to prevent a repeat of its massive overproduction of simulated warm-sector supercells prior to that event, but the pendulum could have swung a little too far in the other direction.