A new method to sparse eigen subspaces by using the pdf-optimized zero-zone quantizers is proposed. It is called sparse Karhunen-Loeve Transform (SKLT). The performance of the proposed method is presented for sparse representation of eigenportfolios generated from empirical correlation matrix of stock returns in NASDAQ-100 index. Performance results show that the proposed SKLT outperforms the popular algorithms to sparse eigen subspaces reported earlier in the literature.