In this appendix we look at how we can reconcile our approach with the Black–Litterman model. Given the subjective nature of both approaches, it is useful to present them using a unified ‘language’. In particular, the Black–Litterman approach can be used (admittedly, in a rather reductive manner) as a regularization device to tame the instabilities in allocations from which the Markowitz model notoriously suffers. We show that the Black-Litterman approach can be married to the Bayesian-net construction that we propose, and that also in this marriage it can play a ‘regularization’ function. In Chapter 29 (see Section 29.2 in particular) we discussed whether this way of achieving a greater stability for the allocation weights is desirable or not.

The Black–Litterman ‘regularization’

We provided in Chapter 9 what we called a somewhat ‘tendentious’ presentation of the Black–Litterman approach – where, when we say ‘tendentious’, we simply mean that we gave a presentation explicitly tailored to highlighting two particular messages out of the many that could be extracted from the approach.

1. (i) The first ‘message’ is that, as the Black–Litterman approach naturally incorporates subjective views into its set-up, it is germane in spirit, if not in implementation, to our view of the world. Therefore combining Black–Litterman with Bayesian nets is a natural logical step. We show in the rest of this chapter how this can be done.