Schweser reports that TVAR equals VAR plus the expected value of losses exceeding VAR. Given that the latter include VAR, how is that not double counting losses?

VAR is minumum loss. TVAR includes loss beyond minumum.

Common sense would tell me TVAR is VAR plus the expected amount by which losses in the tail will exceed VAR

I see your thought process… but…if VAR is 1 and TVAR is 3, 2 remains in the tail beyond VAR. Thus 2+1=3.

Schweser was wrong with its presentation of tail VAR. The CFA curriculum says tail VAR is the conditional tail expectation, so basically it’s the average of cases beyond the VAR point (Volume 5, p. 172).

^+1

TVAR = VAR + AVERAGE (losses in the tail of the distribution - beyond VAR)