Active bond management depends on an economic scenario in order to forecast the movements of yield curve.
A portfolio manager skillfully builds a portfolio with risk exposures that are consistent with his prediction of movements with regard to term structure:
If the portfolio manager forecasts a downward parallel shift in the term structure, then he must ensure whether the managed portfolio is exposed more to the shift factor rather than benchmark ZSp > ZSB
If he forecasts that the slope of the term structure will be steeper or flatter, then in that case the exposure of the managed portfolio to the twist factor must be higher when compared to the benchmark Ztp > ZtB
If he forecasts that the curvature of the term structure changes its level of concavity, then he must make sure that the managed portfolio is more exposed to the butterfly factor than the benchmark.
In case the portfolio manager forecasts that there will be decline in spread of lesser quality bonds, then he must overweight those types of bonds and vice versa and other forecasts.
Thus, a number of ways are available to build bond portfolios along the lines described with regard to equities, such as:
The portfolio manager confirms that the basic statistics such as duration, convexities etc., of managed portfolio are consistent with his scenario.
He can utilize a risk model in order to ensure that he is taking the desired risk exposures.
He can make use of an optimizer to minimize any tracking error of his portfolio, which is subjected to constraints on risk factor exposures that are required to implement his active portfolio.
He can apply a full-fledged optimization technique to maximize the forecasted risk adjusted active return of the bond portfolio while having forecasts for the factor returns.
So, with such an increase in sophistication in building a portfolio, we have to move from a largely judgmental method to a more structured and quantitative method of bond management.