When we derived the showers for AF3 we derived them for electrons and positive hadrons. We should therefore impose a phi -> -phi change for positrons and negative hadrons rather than all negative particles which was previously in the code.
These changes are pretty small but might help some electron ID agreement. Tests of the change on the angle of topoclusters are shown here; https://indico.cern.ch/event/1274181/ indicating that this will be a small change to the physics but is in the correct direction.
Note that this will change the AF3 nightly output.