What kind of effect on target have you seen, chasing SD?
Particularly with OCW-type load development, I've noticed that SD doesn't really have much correlation with group size (I did find better correlation with ES, which makes sense, given that most people don't measure group size as mean radius). As an overall trend, higher charges/pressures produce lower SD, but my velocity loads don't correlate with my accuracy loads.
It's so pronounced in my experience that I might even define an accurate load/node as one which is insensitive to SD. Which implies to me that the negative impact of poor node selection is far more detrimental than the impact of varying velocities.
FWIW: I find my Hornady measures are more accurate with Varget & friends than Dillon.