Qualitative and quantitative information

Another question from LinkedIn - Performance & Risk Analysis group

How do you integrate qualitative and quantitative information in a systematic way?
In mathematics, there exists the Bayesian approach. Related to that, we know the Black/Litterman approach. In medicine and other disciplines, there are "evidence-based" approaches - what is your take on enhancing quantitative with qualitative information in a structured manner?

First, there is no way we can know what is the optimum portfolio. Estimation error means we have a distribution of portfolios that could be optimal.

Secondly: We can use resampling or bootstrapping to give us an idea of this distribution.

Thirdly: Choose a portfolio in this cloud of possible optimums that is line with your qualitative information. By the very nature of qualitative information, this has to be subjective.

Fourthly: Be prepared to have to defend this approach from people who think that it lacks rigour. When I've seen the arguments of such people I find that what is lacking is a rigourous knowledge of the statistical underpinnings of portfolio arithmetic.