District-based elections, in which voters vote for a district representative and those representatives ultimately choose the winner, are vulnerable to gerrymandering, i.e., manipulation of the outcome by changing the location and borders of districts. Many countries aim to limit blatant gerrymandering, and thus we introduce a geographically-based manipulation problem, where voters must vote at the ballot box closest to them. We show that this problem is NP-complete in the worst case. However, we present a greedy algorithm for the problem; testing it both on simulation data as well as on real-world data from the 2015 Israeli and British elections, we show that many parties are potentially able to make themselves victorious using district manipulation. Moreover, we show that the relevant variables here go beyond share of the vote; the form of geographic dispersion also plays a crucial role.