There’s no doubt that little attention has been paid to the necessity of a principal business context, in which the members of an investment committee decide for the optimal synthesis of portfolios, considering several and conflicting investment criteria. The underlying gap widens significantly, considering that mainstream portfolio theory fails to accommodate, both additional investment objectives, beyond expected return and variance, and analysts’ forecasts, when these are coming in the form of discrete recommendations. This post is a summary of a recent publication titled "Grouped data, investment committees & multicriteria portfolio selection", appearing in the Journal of Business Research. The use of humans as wise decision makers, when combined with the wealth of data available to form financial portfolios, is what makes the link of the ideas presented in the paper with the judgmental forecasting literature.

The central contribution of one of our recent papers (Xidonas et al., 2021) is associated with the introduction and standardization of a unified decision support business framework, which deals with all the above complexities and facilitates asset managers in their professional practice. As Elton & Gruber (1987) argue, almost all of the literature in finance analyzing the selection of optimum portfolios assumes that the agent making the decision has a full set of estimates of the expected return for each security and the variance-covariance matrix between securities. However, in actual practice most practitioners simply receive a list of stocks with a ranking on each stock (usually from 1 to 5) and perhaps some partial risk information: if stocks are grouped from 1 to 5, then stocks ranked in group 1 are best buys, while stocks ranked 5 are definite sells. Thus, the practical utilization of grouped data is a critical issue. Since these are the inputs most investors exploit to make investments, the adjustment of classical portfolio theory is necessary in order to secure the optimality of the underlying decisions.

The previously mentioned codings make up the information communicated to an investment committee. However, according to Mottola & Utkus (2009) “the decisions investment committees make, often overshadow an important fact; groups, by their nature, make decisions that are susceptible to behavioral and psychological biases. And these biases may generate irrational decisions, which can influence portfolios’ returns and augment exposure to fiduciary risk”.

Our purpose in Xidonas et al. (2021) was to develop an integrated multicriteria portfolio selection methodology for group decision support, combining the business practice of grouped data with the classical theory of portfolio analysis; moreover, the value of the suggested framework is enriched with the introduction of a novel concept in the field of investment decision making for groups, i.e., the ‘signal-consensus impact plane’, a data analytics tool which provides effective visualization for the decision makers. The suggested framework fundamentally recognizes that the typical portfolio selection decision process is associated with five (5) problem dimensions and an equal number of solution components:

a) Grouped data,

b) Investment committees,

c) Investment objectives,

d) Portfolio theory, and

e) Approved lists.

Each one is confronted with a specific modeling tool from the field of operations research and decision sciences (Figure 1). The underlying platform, a fully tractable routine, is also adequately compatible with the current normative trends in the whole spectrum of investment management processes, towards highly controlled operations, consistent with the notion of compliance in capital markets.