Multicriteria portfolio management

What this book actually teaches

Most portfolio optimization textbooks treat investment as a single-objective problem: maximize return for a given level of risk, or minimize risk for a target return. Panos Xidonas's "Multicriteria Portfolio Management" challenges that framing by arguing that real portfolio construction involves multiple, often conflicting objectives — return, risk, liquidity, ESG factors, transaction costs, tracking error — that cannot be collapsed into a single utility function without distorting the decision.

The book applies Multiple Criteria Decision Analysis (MCDA) methods to portfolio construction. Where classical Markowitz mean-variance optimization asks investors to specify a single risk tolerance and returns one efficient frontier, multicriteria approaches allow decision-makers to specify preferences across several dimensions simultaneously and evaluate trade-offs explicitly. Xidonas covers a range of MCDA techniques — including goal programming, PROMETHEE, ELECTRE, and multi-objective evolutionary algorithms — and shows how each can be applied to asset selection and portfolio construction problems.

The empirical sections are grounded in real market data, primarily from European equity markets, and demonstrate that multicriteria methods can identify portfolios that dominate single-criterion solutions when the decision-maker's true preferences span more than one dimension. The argument is not that mean-variance is wrong in theory but that it is wrong for practice, where investment mandates routinely impose constraints — maximum sector concentration, minimum dividend yield, ESG screens — that a single-objective model handles poorly.

This is an advanced academic text aimed at quantitative analysts, doctoral students in finance or operations research, and institutional portfolio managers with strong quantitative backgrounds. Readers without familiarity with mathematical programming, linear algebra, and portfolio theory will find it inaccessible.

The main limitations are practical rather than theoretical. The MCDA methods the book describes require preference elicitation from decision-makers — investors must articulate their trade-offs explicitly — which is harder in practice than it sounds. The book is also more concerned with the mechanics of the methods than with implementation in production environments, and the empirical examples, while illustrative, predate significant market structure changes. The coverage of computational complexity and scalability to large universes is lighter than practitioners would want.

For a quantitative researcher or academic working on portfolio optimization beyond the mean-variance framework, this is a systematic and technically rigorous reference. For practitioners, it is more useful as a source of conceptual frameworks than as an implementation guide.

Read next

About Panos Xidonas

Read more from Panos Xidonas and explore the full bibliography on ClearValue Books.