ABSTRACT We perform an analysis of two-point galaxy clustering and galaxy bias using Subaru Hyper-Suprime Cam (HSC) data taken jointly by the Subaru Strategic Program and the University of Hawaii in the Cosmic Evolution Survey (COSMOS) field over an area of 1.8 sq deg. The depth of the data is similar to the ongoing Hawaii Two-0 (H20) optical galaxy survey, thus the results are indicative of future constraints from tenfold area. We measure the angular autopower spectra of the galaxy overdensity in three redshift bins, defined by dropouts from the g, r, and i bands, and compare them to the theoretical expectation from concordance cosmology with linear galaxy bias. We determine the redshift distribution of each bin using a standard template-based photometric redshift method, coupled with a self-organizing map to quantify colour space coverage. We also investigate sources of systematic errors to inform the methodology and requirements for H20. The linear galaxy bias fit results are $b_{\mathrm{gal,g}} = 3.90 \pm 0.33 (\mathrm{stat}) \substack{ +0.64 \\ -0.24 } (\mathrm{sys})$ at redshift z ≃ 3.7, $b_{\mathrm{gal,r}} = 8.44 \pm 0.63 (\mathrm{stat}) \substack{ +1.42 \\ -0.72 } (\mathrm{sys})$ at z ≃ 4.7, and $b_{\mathrm{gal,i}} = 11.94 \pm 2.24 (\mathrm{stat}) \substack{ +1.82 \\ -1.27 } (\mathrm{sys})$ at z ≃ 5.9.