Mean squared error
Also known as: MSE, Quadratic risk
Mean squared error of an estimator $W$ for $\theta$ is
\[\mathrm{MSE}_\theta(W)=E_\theta[(W-\theta)^2]=\mathrm{Var}_\theta(W)+\mathrm{Bias}_\theta(W)^2.\]It scores overall accuracy: variance and squared bias. Unbiased ⇒ MSE $=$ Var. Bias–variance trade-offs are common (e.g. Normal variance MLE vs $S^2$ at small $n$).