A sample ratio mismatch (SRM) is a gap between the observed split of units across groups and the expected one, for example 50/50. It signals an assignment or data collection problem and makes the results unreliable. It is detected with a chi-square test on observed and expected counts.
Example
from scipy.stats import chisquare
chisquare([5200, 4800], f_exp=[5000, 5000])