What is the optimal share for solar-wind hybrid projects regarding grid reliability? What are the metrics and time scales of concern? An increasing number of studies are assessing this subject from distinct approaches. This work aims at unfolding these questions for a specific site in Northeastern Brazil where wind and solar power projects are already profitable. In Brazil, wind responds actually for the third largest source of electrical power, accounting for ~ 9% of annual production. Solar power is experiencing high growth rates in the last years confirming the scenarios that it will become the fourth largest source by 2030. Despite the large share of hydroelectricity that could absorb short-term wind and solar variability, integrating both sources in a way to minimize impacts is a key factor to improve grid reliability. In this context, this study evaluates the local complementarity of wind-solar hybrid projects from different perspectives, assessing metrics and time scales that could infer the optimal share of both sources. Ground data for solar irradiance and 100-meter wind speeds is used to feed a wind and solar power conversion model at 30-minute resolution for a 14-year period. Capacity factors (CF) time series of both sources are correlated at different timescales for a complementarity analysis. Combined CFs for different shares, namely Solar Fractions (SF) are evaluated for several metrics. Metrics varies such as minimum standard deviation (SD), maximum sustained monthly production for a year-long or minimum number of days below specific power threshold. It is observed that solar power presents a much lower variability on all but hourly time scales when compared to wind power. This is not surprisingly due to nature of insolation at tropical regions. Average solar CF varies between 0.15 and 0.20 while average wind CF varies between 0.15 to 0.35. The observed lag between wind and solar CF cycles are around 2 months showing weak complementary pattern on the seasonal scale. The optimum mix depend on the criteria, ranging from a SF of 0.7 that maximizes the minimum monthly power production, to a SF of 0.9 that allows the lowest seasonal standard deviation. This study present a novel approach in the region to the study of the complementarity of solar and wind energy sources in different scales and considering a variety of metrics, and used a comparatively large temporal series of observational solar and wind data. Preliminary results show that the optimum share for solar-wind hybrid projects will depend on the scale and metrics adopted. Although minimization of SD is regularly used for grid optimization, SD is lower when CF is lower, so this can be a good proxy for grid reliability, since SD is proportional to power and thus comparable between months. But this is not true when investors are looking for minimum revenue risks at monthly scales, require to re-scale SD by average monthly production, leading to different optimum solar fractions. Acknowledgements: This work is a contribution of the project 1042118005 - NUBI through ANEEL R&D program funded by CENPES/PETROBRAS.