Upon reading headlines like “Traffic Fatalities Fell Last Year,” people often overestimate how well they would have anticipated changes. In the present study, participants who estimated fatality statistics and listed possible causes before learning true statistics (Foresight) were more surprised than those who listed possible causes only after learning true statistics (Hindsight). Pezzo (2003) linked hindsight bias to causal explanations that minimize initial surprise after an outcome—but to what extent can people build an expectation for alternative causation before learning the true outcome? Before seeing true statistics, a subset of our Foresight participants listed causes for changes in the opposite of their expected direction. This did not reduce surprise for those who learned that statistics had moved in the opposite of expected direction, but the number of opposite causes they provided reliably predicted their second set of estimates. Moreover, participants frequently explained surprising statistics using their earlier alternative causes.