The discretionary wealth approach derives an appropriate risk aversion parameter from an extended balance sheet including present values of future financial commitments and sources of contribution to the investment portfolio.  It assumes the goal is long-run maximization of median wealth without hitting intermediate shortfall points.  In this, it supplies the missing risk aversion parameter for conventional Markowitz mean-variance optimization.  But what if we instead marry this approach with portfolio optimization according to fully-Bayesian logic?  The ensuing exploration illuminates the portfolio construction problem and appears to provide better answers for difficult cases such as long-short portfolios.