2025MNRAS.542.1705A

New, ultra-deep medium-width photometric coverage with James Webb Space Telescope (JWST)'s NIRCam instrument provides the potential for much improved photo-z reliability at high redshifts. In this study, we conduct a systematic analysis of the JADES Origins Field, which contains 14 broad- and medium-width near-infrared bands, to assess the benefits of medium band photometry on high-z completeness and contamination rates. Using imaging reaching AB mag $29.8\!-\!30.35$ depth, we test how high-z selections differ when images are artificially degraded or bands are removed. In parallel, the same experiments are conducted on simulated catalogues from the JAGUAR semi-analytic model to examine if observations can be replicated. We find sample completeness is high ($80~{{\rm per\,cent}}+$) and contamination low ($\lt 4~{{\rm per\,cent}}$) when in the $10\sigma +$ regime, even without the use of any medium-width bands. The addition of medium-width bands leads to increases in completeness ($\sim 10~{{\rm per\,cent}}$), but multiple bands are required to improve contamination rates due to the small redshift ranges over which they probe strong emission lines. Incidents of Balmer-Lyman degeneracy increase in the $5{\!-\!}7\sigma$ regime and this can be replicated in both simulated catalogues and degraded real data. We measure the faint-end of the ultraviolet luminosity function (UV LF) at $8.5\lt z\lt 13.5$, finding high number densities that agree with previous JWST observations. Overall, medium bands are effective at increasing completeness and reducing contamination, but investment in achieving comparable depths in the blue ($\lt 1.5\,\mu$m) as achieved in the red is also found to be key to fully reducing contamination from high-z samples.