GNSS-IR is a method that enables retrieving characteristics of the reflection surface by analyzing the interference between direct and coherently reflected signals transmitted from satellites. The selection of azimuth and elevation angles is of great importance in this method. Traditional methods for determining azimuth and elevation angle masks may be insufficient due to irregularities and time-dependent changes on the reflection surface. In this study, we propose a novel empirical approach that allows determination of optimal azimuth and elevation angle masks. The approach is based on the RMSE and correlation values of estimates obtained by directly evaluating the SNR data with in situ measurements. The proposed approach was trained with four different strategies (subsequently three additional strategies) using 1-week GPS L1C SNR data of the GADA station located in the Aegean Sea and co-located with a tide gauge covering the period December 13–19, 2020. After identifying the angle masks, GNSS-IR analysis of 6-month data from the same station between January 1 and June 30, 2021 was performed using the angle masks found for each strategy combination, and then daily and sub-daily (for 12-h, 6-h, 4-h, 3-h, 2-h, and 1-h intervals) sea level estimations were obtained. Compared with tide gauge measurements, daily estimations showed a correlation of over 97% and an RMSE below 3.2 cm. For sub-daily estimations, a correlation of over 85% and an RMSE below 10 cm were obtained even for the 1-h estimation interval. The results show that the approach proposed here can be used to determine optimal azimuth and elevation angle masks in GNSS-IR analysis without requiring any data other than in situ measurements during the pre-analysis period.