Wrong neutron elastic scatter angular distribution

Hi, experts:

The angular distribution for 0.1 MeV neutron elastic scatter in Fe-56 (only accumulate the first step to guarantee the accurate energy) is not similar with nuclear data.

Here’s the distribution for ENDF/B 7.1:
output

Here’s the distribution for Geant4 v11.0.1:
cosine

Though the figure from ENDF/B 7.1 is probability and one from Geant4 is bins and count, they should have the similar shape. They are obviously different in the results. Why?

EDITED: Updated figure for ENDF/B 7.1 because I forgot to convert it into lab coordinate.

Update: The result for HP physical process. Much better, but not same with too.

cosine