The definition of “race” you use is genetically invalid. But populations do have validity in genetics- so this is the evidence that I would accept:
First, an identification of all the genes that are involved in intelligence (or at least the most important genes) and propensity for criminality, and exactly how these genes are involved (which ones of them are active when “turned on” or combined with specific other genes, and so on), and how they relate to environment and upbringing. Then detailed, broad genetic testing of various populations (not races), showing that certain populations have significantly varying numbers of these genes. Then detailed, broad surveys and testing that shows that the previous analyses have predictive power.
This is genetic evidence. You’ve supplied zero genetic evidence- not even a single gene. Real science is hard.