OFDM systems have gathered quite some attention from international standardization bodies for the past decade. Various works in literature aim at optimizing transmission for these systems in the open-loop scenario, i.e. when there is no feedback communication between transmitter and receiver. This work focuses on the utilization of a closed-loop for OFDM systems, a subject which has not been thoroughly explored so far. Here we primarily focus on the recently proposed method of dynamic pilot-symbol allocation. It was shown in a recent paper that for coherent closed-loop OFDM systems, this option is capable of delivering substantial increases in performance. The pilot-symbols are used to sample channel behavior so that data symbols can be decoded correctly. However, these pilot do not need to be located so as to estimate the channel in the best possible way. The system requires an estimation which is only good enough. In this work, we propose a pilotsymbol allocation strategy with the goal of minimizing the bit error rate. We also suggest a lower complexity algorithm, which attains most of the performance of the optimal solution. Experimental results show significant gains.