Unlike traditional valuation methods, viability theory provides tools for eradicating the risk, by determining the minimum initial capital that would meet the commitments of the investor, regardless market developments. In this study, we compare two approaches to risk assessment within a framework of asset-liability management (ALM) of a guaranteed fund. The optimal allocation of assets for such funds is determined initially by the classical portfolio insurance (thus with a statistical evaluation of risk) and then in a second step, by tools of viability theory. Although the results from the two approaches are not strictly comparable in terms of numerical point of view (as in both cases the goals are different in nature), this study offers, on a practical example of ALM management, two radically different philosophies: one is the statistical evaluation of risk, based on probabilistic models, while the other one eradicates risk, using viability theory.