The caustic technique for measuring mass profiles of galaxy clusters relies on the assumption of spherical symmetry .
When applied to aspherical galaxy clusters , the method yields mass estimates affected by the cluster orientation .
Here we employ mock redshift catalogues generated from cosmological simulations to study the effect of clusters intrinsic shape and surrounding filamentary structures on the caustic mass estimates .
To this end , we develop a new method for removing perturbations from large-scale structures , modelled as the two-halo term , in a caustic analysis of stacked cluster data .

We find that the cluster masses inferred from kinematical data of ~ { } 10 ^ { 14 } M _ { \hbox { $ \odot$ } } clusters observed along the major axis are larger than masses from those observed along the minor axis by a factor of 1.7 within the virial radius , increasing to 1.8 within three virial radii .
This discrepancy increases by 20 \% for the most massive clusters .
In addition a smaller but still significant mass discrepancy arises when filamentary structures are present near a galaxy cluster .

We find that the mean cluster mass from random sightlines is unbiased at all radii and their scatter ranges from 0.14 to 0.17 within one and three virial radii , with a 40 \% increase for the most massive clusters .
We provide tables which estimate the caustic mass bias given observational constraints on the cluster orientation .