Results for the CAMIRA-like cluster-finding algorithm in our mock catalog. Left-hand panel: Comparison of estimated locations of |$r_{\rm sp}^{\rm 3D}$| based on the DK14 profile fitting assuming the spherical symmetry to the projected cross-correlation functions between halos and the HOD galaxies in the simulation, and those from the emulator stack when selecting halos based on halo masses (i.e., without projection effects related to richness values). As expected from the construction of the HOD galaxy catalog, the two values match quite well (see text for more details). The blue horizontal line denotes the average value of the ratio, 1.05. We estimate the error-bars from the MCMC chains with half widths of the |$68\%$| percentile region. Right-hand panel: Similarly to the left-hand panel, we compare the observed splashback radii from the DK14 fitting to the projected cross-correlation functions with those from the emulator stack, but here we select halos based on richness values with projection effects. We use 10 richness bins with bin edges of N = [15.0,17.6,20.6,24.2,28.4,33.3,39.1,45.9,53.8,74.0,200.0]. The constraint for the CAMIRA-like finder is |$r_{\rm sp, DK14\ MCMC}^{\rm 3D}/r_{\rm sp, emulator\ stack}^{\rm 3D} = 1.06 \pm 0.03$| for the sample of 15 < N < 200, which is consistent with 1.05 from the mass selection in the left-hand panel without projection effects, and its uncertainty is smaller than our statistical errors in the HSC data analysis. Note that we estimate this ratio by repeating the procedure for the cluster sample of 15 < N < 200. (Color online)