Previously, we reported an electron-water pseudopotential designed to be used in conjunction with a polarizable water model, in order to describe the hydrated electron [L. D. Jacobson et al., J. Chem. Phys. 130, 124115 (2009)]. Subsequently, we found this model to be inadequate for the aqueous electron in bulk water, and here we report a reparametrization of the model. Unlike the previous model, the current version is not fit directly to any observables; rather, we use an ab initio exchange-correlation potential, along with a repulsive potential that is fit to reproduce the density maximum of the excess electron's wave function within the static-exchange approximation. The new parametrization performs at least as well as the previous model, as compared to ab initio benchmarks for (H(2)O)(n) (-) clusters, and also predicts reasonable values for the diffusion coefficient, radius of gyration, and absorption maximum of the bulk species. The new model predicts a vertical electron binding energy of 3.7 eV in bulk water, which is 1.4 eV smaller than the value obtained using nonpolarizable models; the difference represents the solvent's electronic reorganization energy following electron detachment. We find that the electron's first solvation shell is quite loose, which may be responsible for the electron's large, positive entropy of hydration. Many-body polarization alters the electronic absorption line shape in a qualitative way, giving rise to a high-energy tail that is observed experimentally but is absent in previous simulations. In our model, this feature arises from spatially diffuse excited states that are bound only by electronic reorganization (i.e., solvent polarization) following electronic excitation.