Floating-tray hydroponics is expanding for microgreen production, yet evidence on substrate performance under natural-light greenhouses remains limited. This study compared perlite, coconut coir, and a 1:1 (v/v) perlite–coconut coir mixture for amaranth (Amaranthus cv. ‘Diablo Rojo’) grown at the end of summer in an unheated greenhouse (CRD; 4 replicates). Perlite significantly improved establishment and yield: emergence reached 72.6% versus 46.1% in coconut coir and 35.1% in the mixture, with fresh biomass of 297, 171, and 119 g m⁻², respectively. The proximate composition exhibited consistency across substrates (protein ≈ 32% DW; crude fiber ≈ 32% DW; crude fat ≈ 2.5% DW), whereas nitrate concentrations varied between 1300 and 2500 mg kg⁻¹ FW, irrespective of the substrate. Vitamin C and total phenolics showed no significant variation, and β-carotene exhibited only a slight, non-significant increase in coconut coir. In contrast, mineral composition was substrate-dependent: perlite increased Ca (3626 ppm DW) and Mg (1094 ppm DW), while P and Fe were unaffected; Na was higher in perlite. These results indicate that under natural-light, unheated greenhouses, bioactive compounds are mainly influenced by environmental variability, whereas root-zone aeration drives Ca/Mg enrichment and yield. Perlite-based floating-tray systems represent efficient, low-cost strategies with potentially lower environmental impact, suitable for scalable urban and smallholder microgreen production. These findings, derived from a single-season trial in one unheated greenhouse, should be regarded as preliminary yet consistent with previous studies on microgreens. Future multi-season and multi-species experiments will help to confirm and expand on these results.