So at my annual review in December I should only be assessed on work in the last quarter? I guess I'd better save my best work for November.

Depending on the attention span of your particular managers (and their managers as well, most likely), this can indeed be a viable strategy.

You might also carefully consider your company's culture when deciding how to split your efforts between new (user-visible) features, preventive maintenance, and bugs affecting end users (especially any VIP end users that have your managers' ears).

For example, a 'Hero' culture rewards last-minute, late night, death march efforts, whether to fix an urgent problem, or to add that one feature that is (supposedly) needed to land a specific high-value customer/sale.