The CALPHAD method allows the prediction of metastable equilibria. However, the predictive power of metastable phase equilibria has limitations since the phase description itself relies on experimental data of a system in thermodynamic equilibrium only.  As a workaround, the solidification community uses estimated phase diagrams, which have the necessary reliability in the metastable range, but lack the consistency of the CALPHAD method that allows the  calculation of derived quantities such as free energy, enthalpy or entropy. In the presentation a new strategy is presented that uses experimental and simulation results for optimizing phase equilibria at high undercoolings.