Choosing a control group is vital to quasi-experimental design in the social sciences. When randomized trials are not feasible or ethical, researchers (usually) need to compare a treated unit/set of units to others that were untreated. To do this, fdid uses a greedy forward selection algorithm to choose the ideal control group for the currently treated unit. It then uses the standard difference-in-differences model to estimate effects. The fdid command reports a full suite of inferential statistics, metrics of fit, as well as the control units used for the difference-in-differences design.