I have shot over 300 groups each in 9x19, .40S&W, and .45Auto. In each case, they were matched between cases that had the SAME headstamp and as close to the same weight a possible compared to a grab bag of random never-heard-of headstamps.
Averaging through all this data, for each caliber, the mixed brass tended to have an average smaller group size by about 0.20-0.45" in all three cartridges.
Comparing those cases where the matched cases produced a smaller group and where the mixed (and I do mean mixed) produces a smaller group, the mixed just edged out the matched about 52% to 48%.
Each matched pair was fired through the same gun, randomly, without my knowledge of what case and charge weight I was shooting--as blind as I could easily make it.
I say--It just doesn't matter.