The hybrid of spatial multiplexing and transmit beamforming is a simple scheme to combat channel fading in multiple-input multiple-output (MIMO) transmission. In this scheme, a set of transmit antennas is used to transmit a bit stream and antenna grouping is then required. How to conduct the grouping optimally is the main concern. For maximum-likelihood (ML) receivers, the grouping criterion is equivalent to maximizing the minimum distance of receive signal constellations, referred to as free distance. However, finding the free distance needs an exhaustive search which is not desirable. In the literature, a singular-value-decomposition (SVD) based method was proposed to obtain a lower bound of the free distance. Using the lower bound as the grouping criterion, one can solve the problem much easier. In this paper, we first propose using a QR decomposition (QRD) based method to obtain another lower bound. It can be shown that the lower bound yielded by the QRD is tighter than that by the SVD. We then propose using a basis-transformation method such that the QRD-based lower bound can be further tightened. Simulations show that the basis-transformed QRD-based method can yield near-optimum performance.