The large crescents imaged by ALMA in transition disks suggest that azimuthal dust trapping concentrates the larger grains, but centimetre-wavelengths continuum observations are required to map the distribution of the largest observable grains. A previous detection at ∼1 cm of an unresolved clump along the outer ring of MWC 758 (Clump 1), and buried inside more extended sub-mm continuum, motivates followup VLA observations. Deep multiconfiguration integrations reveal the morphology of Clump 1 and additional cm-wave components which we characterize via comparison with a deconvolution of recent 342 GHz data (∼1 mm). Clump 1, which concentrates ∼ 1/3 of the whole disk flux density at ∼1 cm, is resolved as a narrow arc with a deprojected aspect ratio χ > 5.6, and with half the azimuthal width than at 342 GHz. The spectral trends in the morphology of Clump 1 are quantitatively consistent with the Lyra-Lin prescriptions for dust trapping in an anticyclonic vortex, provided with porous grains ( f ∼ 0.2 ± 0.2) in a very elongated ( χ ∼ 14 ± 3) and cold (T ∼ 23 ± 2 K) vortex. The same prescriptions constrain the turbulence parameter α and the gas surface density Σ g through log 10 α × Σ g /g cm −2 ∼ −2.3 ± 0.4, thus requiring values for Σ g larger than a factor of a few compared to that reported in the literature from the CO isotopologues, if α 10 −3 . Such physical conditions imply an appreciably optically thick continuum even at cm-wavelengths (τ 33 GHz ∼ 0.2). A secondary and shallower peak at 342 GHz is about twice fainter relative to Clump 1 at 33 GHz. Clump 2 appears to be less efficient at trapping large grains.