By means of return-based style analysis (RBSA), heterogeneous style sub-categories were identified within the TARR category of the South African unit trust market to create a framework for sub-categorisation. The study dealt with TARR funds and their place within the investment universe. The literature review emphasised the importance of asset allocation, which supports the use of RBSA to identify asset allocation. The literature review further provided a motivation for the semi-strong form of RBSA applied to the sample data. In the study, RBSA was applied to two groups within the sample data, namely funds that have data points for the full measurement period (Group 1) and funds that have less than 75 data points (Group 2). A four-phase process was applied to the sample data. The findings suggest the following:<ul><li> in general, return-based style analysis applied to each fund identifies the asset allocation for the fund and is valid; but it is emphasised that for specific periods, the explanatory power of the regression model may become questionable; </li><li> the collective results of return-based style analysis applied to the funds can be used to create a framework for sub-categorisation. The framework proposed was the result of nine out of a potential 54 funds. The explanatory power of the regression results was less questionable. The proposed framework was applied to the remaining 45 funds (Group 2), but there were indeed inconsistencies in the application; </li><li> the framework created did not raise any concerns as a result of the Group 1 analysis. However, it was questionable when applied to the Group 2 funds in its entirety; </li><li> sub-categorisation based on only the allocation to the domestic short-term asset class was definitely a criterion that was true irrelevant of which sample group it was applied to. </li></ul> / Dissertation (MCom)--University of Pretoria, 2011. / Financial Management / unrestricted