This paper undertakes a holistic investigation of two fundamental trade-offs in monostatic OFDM integrated sensing and communication (ISAC) systems-namely, the time-frequency trade-off and the spatial trade-off, originating from the choice of modulation order for random data and the design of beamforming strategies, respectively. To counteract the elevated side-lobe levels induced by varying-amplitude data in high-order QAM signaling, we propose a novel linear minimum mean-squared-error (LMMSE) estimator, capable of maintaining robust sensing performance across a wide range of SNRs. Moreover, we explore spatial domain trade-offs through two ISAC transmission strategies: concurrent, employing joint beams, and time-sharing, using separate, time-non-overlapping beams for sensing and communications. Simulations demonstrate superior performance of the LMMSE estimator, especially in detecting weak targets in the presence of strong ones with high-order QAM, consistently yielding more favorable ISAC trade-offs than existing baselines under various modulation schemes, SNR conditions, RCS levels and transmission strategies. We also provide experimental results to validate the effectiveness of the LMMSE estimator in reducing side-lobe levels, based on real-world measurements.