Truncating an N-bit signed by an integer is usually an N-bit signed. (Weak form, enabled by default).
See also basic-signed-byte-p-of-truncate-split.
Theorem:
(defthm basic-signed-byte-p-of-truncate (implies (and (signed-byte-p n a) (integerp b) (not (and (equal a (- (expt 2 (- n 1)))) (equal b -1)))) (signed-byte-p n (truncate a b))))