A proprietary open-source simulation and optimization framework, named LMR, has been used to model LMR sensors with cylindrical geometry. Particular attention is given to the analysis of two fitness functions used in parameter optimization: the commonly applied Mean Squared Error (MSE) and a custom-designed metric, the Valley Error (VE). The VE metric has been demonstrated to be more suitable for capturing spectral features essential to LMR sensor performance, such as the position and shape of the resonance dip. Simulation studies involving TiO2-based coatings with varying thicknesses highlight the practical advantages of using the VE metric, especially in cases where a precise determination of the refractive index and the extinction coefficient is required without prior knowledge of the layer thickness. Compared to MSE, VE provides more accurate and functionally relevant optimization results. These findings indicate that LMR sensors, when combined with appropriately selected optimization criteria, can serve not only as sensing devices, but also as effective tools for material characterization, offering a potential alternative to ellipsometry. Furthermore, the results reveal that similar spectral responses can be achieved using geometrically distinct sensor designs, suggesting new opportunities for performance tuning through structural optimization.