Election outcomes in Western democracies are increasingly difficult to predict since vote intentions now often shift quickly during the campaign period. Standard theories of electoral behavior cannot account for these quick changes in vote intentions before an election. We investigate whether strategic coalition voting and bandwagoning theory can explain variance in vote intentions in the short run. We analyze it in a set of ten elections in Spain. We test the theories based on original data from a real money prediction market, which has been designed in order to measure dynamics in expectations about the likelihood of the formation of alternative governments and the party affiliation of the winning president. We relate these vote intentions to a series of poll-of-polls and find both evidence for strategic coalition and bandwagoning. We elaborate on the conditions that determine the activation of both mechanisms. We argue that our dynamic approach with aggregate–level data is complementary to individual-level analyses and will allow the literature on strategic coalition voting and bandwagoning to move from case studies to systematic comparisons.