ABSTRACT The study of stellar flares has increased with new observations from CoRoT, Kepler, and TESS satellites, revealing the broad-band visible emission from these events. Typically, stellar flares have been modelled as 104 K blackbody plasma to obtain estimates of their total energy. In the Sun, white-light flares (WLFs) are much fainter than their stellar counterparts, and normally can only be detected via spatially resolved observations. Identifying the radiation mechanism for the formation of the visible spectrum from solar and stellar flares is crucial to understand the energy transfer processes during these events, but spectral data for WLFs are relatively rare, and insufficient to remove the ambiguity of their origin: photospheric blackbody radiation and/or Paschen continuum from hydrogen recombination in the chromosphere. We employed an analytical solution for the recombination continuum of hydrogen instead of the typically assumed 104 K blackbody spectrum to study the energy of stellar flares and infer their fractional area coverage. We investigated 37 events from Kepler-411 and five events from Kepler-396, using both radiation mechanisms. We find that estimates for the total flare energy from the H recombination spectrum are about an order of magnitude lower than the values obtained from the blackbody radiation. Given the known energy transfer processes in flares, we argue that the former is a physically more plausible model than the latter to explain the origin of the broad-band optical emission from flares.