I just stumbled across an interesting chart in Meir Statman's book "Finance for Normal People" where he introduces his behavioral portfolio theory. There, he also provides the following chart which I found to be quite interesting:
I understand from his other papers that this kind of efficient frontier can be derived from the safety-first approach. How can I construct this behavioral efficient frontier for a set of various behavioral biases (e.g. correlation neglect, overconfidence etc.)?
My interpretation is that a behavioral efficient frontier provides all portfolio combinations that maximize the portfolio expected return s.t. a given level of behavioral bias quantified e.g. by a transformation of the covariance matrix or by a modified expected return vector.
My approach so far concerning correlation neglect in the sense of Eyster and Weizsäcker (2016) is to maximize expected returns s.t. that the behavioral covariance matrix equals some variance level.
Any ideas on how to formalize this? Under which conditions are these behavioral portfolios feasible (i.e. the behavioral efficient frontier is located below or at the usual (Markowitz) efficient frontier)?