This paper considers the optimal design of dynamic research contests when the buyer can set time-dependent prizes. We derive the buyer-optimal contest and show that it entails an increasing prize schedule. Remarkably, this allows the buyer to implement a global stopping rule. In particular, the optimal contest attains the first-best. More generally, we show that global stopping rules can be implemented robustly and compare them to individual stopping rules which have been analyzed in the existing literature. We conclude by discussing policy implications of our findings and highlight that global stopping rules combine the best aspects of innovation races and research tournaments.