In hermaphroditic dichogamous species, sporophylls are separated in time. However, the separation of sporophylls can lead to pollination inaccuracy, especially in movement-assisted dichogamy, where sporophylls alter their position over time. Is pollination inaccuracy minimised by the second sporophyll taking the exact position of the first? We address this question in Clerodendrum infortunatum, a movement-assisted dichogamous species. We made predictions from optimality arguments, and tested these by measuring sporophyll angles over time, by experimentally manipulating flowers, and by estimating correlates of the resultant fitness, taking into account pollen export, pollination inaccuracy and the resultant total pollen delivered. Contrary to expectation, anthers do not have a fixed position in the male phase, and pollination inaccuracy is high. Further, when the number of pollen grains on anthers is highest, anthers are paradoxically not positioned at the pollen export peak. Also, pollen export and pollen deposition peaks do not align. Despite this seeming maladaptiveness, the starting position of the anthers is, in fact, very close to the theoretical optimum, which results in the highest overall male fitness, measured as the total pollen delivered over the entire male phase. Instead of a simple positional exchange of sporophylls, stamens display a more complicated dynamic strategy that appears close to optimal even though naive measures of pollination inaccuracy are high.