Suppose that $X$ and $B$ are two random variables that are correlated. Then for any $\alpha$ and $\lambda$ real numbers who belong in the $\mathbb{R}-\{0\}$ holds
$$\tag{1}\mathbb{V}ar(\alpha X-\lambda B)=\alpha^2\mathbb{V}ar(X)-2\alpha\lambda\mathbb{C}ov(X,B)+\lambda^2\mathbb{V}ar(B)$$
what does the expression mean ``The above quantity (1) is is constant-invariant"?
If $X'=X+c$ and $B'=B+d$ where $c$ and $d$ are constants then:$$\mathsf{Var}(\alpha X'-\lambda B')=\mathsf{Var}(\alpha X-\lambda B)$$