An important aspect of animal improvement is the correct identification of those individuals with superior genetic merit. Testing to accurately assess merit, however, can be a costly component of improvement. In the Group Nucleus there is a substantial number of potential offspring to be tested even within a system where the farrowing and testing is practised continuously throughout the year. In order, therefore, to reduce costs it could be beneficial to accurately predict genetic worth prior to test on either mean parental breeding value or a trait expressed early before test.

A Group Nucleus population was modelled by stochastic simulation with selection at two stages The model for the trait(s) selected comprised animal, common environment, herd-year-season and error components. A multivariate animal model BLUP routine was used for the evaluation.