Asset Allocation

Alexander Denev

The application of Bayesian nets to asset allocation has already been extensively explored in Rebonato and Denev (2014). In this chapter, while summarising the key points of the previous work, other representations, such as efficient frontiers, which were not investigated in Rebonato and Denev (2014) and with which certain institutions may be more familiar, will be explored. We give an application to a UK pension scheme where the liability side of the institution, and not just the assets, must be taken into account in the asset allocation process.

We must stress at this point that PGMs allow us to model events that are not necessarily extreme from a forward-looking stance. Whenever there is a potential change with respect to what the time series actually suggest, eg, a structural break, PGMs allow us to organise and make consistent our knowledge around this.

5.1 BACKGROUND

In the late 1950s Markowitz laid down the foundations of the modern asset-allocation approach (see Markowitz 1987, 1991) through a new theory (modern portfolio theory (MPT)), which has since become one of the main pillars of investment decisions. Although it has its limitations, the main idea of