According to Stanford, the goal of Chebyshev's inequality is to bound the probability that the variable is far from its mean (in either direction).