Looking at FBI "violent crime" statistics for the years 2010-2015, 13% crossed racial boundaries...87% of violent crimes were "same-on-same".
Black on black was slightly higher at 92% while white on white was slightly lower at 84%.
Interestingly, when you consider the differential crime levels between urban and rural areas (lower per capita in rural) and the racial population distributions urban vs. rural (urban higher black nationwide except in Mississippi, Alabama, and one other southern state I can't remember)...the difference between black on black vs. white on white is virtually nonexistent.
It can always be argued that the urban vs. rural difference is an effect of the racial distribution (i.e more crime in the cities because higher AA population) and there's nothing in the data that proves/disproves to that point . Most believe that population density is the independent variable and the crime rate is dependent on the density meaning that crime rates in urban areas are driven by people just being more packed in vs. rural areas. And, since those people packed in skew slightly higher AA, that where you get the higher black-on-black average numbers.
Add to that the fact that most violent crimes occur within families or acquaintance groups and they tend to be single race and you get the high same-on-same rate.
Personally, I'd argue that there's not a lot of difference between 84 and 92 percent given all the other noise that is part of this.
I'm sure there's other data that debunks this "black-on-black" chit but this was what I found in my reading.