Prefblog reader Drew asks on the Comments and Requests: HIMIPref thread:
My impression is that HIMIPref does not merely select the most attractive issues for purchase but rather factors in the relative risks between different categories of preferred shares – chiefly retractible versus perpetual – with a view to constructing a safer portfolio than if the focus was placed solely on the most attractive issues. Is my impression correct?
This impression is essentially correct. A lot of thought has gone into the risk control aspects of HIMIPref, although the exercise of these controls is much more explicit when optimizing according to the portfolio method than with the issue method.
With the parameterization currently supplied with HIMIPref for the issue method, a great deal of emphasis is placed on the potential for trading profits and the value of each issue relative to its price given all its risk attributes. Thus, the primary comparison for a split-share, for example, is other split shares; comparisons to perpetuals have a great influence, but are less important. It is unlikely that any class of shares could have so many issues far above the curve that they totally dominate other classes … after all, if there were too many issues like this, then the curve itself would move!
Thus, a certain amount of diversification of risk-attributes is built into the trading model. Should the subscriber wish it, this diversification can be amplified or reduced, by fiddling with the constraintSpecificationRecord that resides on the client-side.