The integration of meteorological, edaphic, and genetic data with robust analyses such as machine learning and factorial regression helps clarify the factors related to high soybean productivity. This study was developed in order to identify the key management and edaphoclimatic factors determining Brazil’s top-performing soybean farmers. Data were collected from the Brazilian Soy Strategic Committee (CESB) website, covering 50 farmers from 36 environments between 2014 and 2023. A total of 18 top-performing cultivars were identified, with relative maturity groups ranging from 5.4 to 8.3. Grain yield was analyzed using centered means with partial least squares, followed by linear regression models and t-tests (p < 0.05). Reaction norm parameters were estimated via the Finlay-Wilkinson method, stratified by production region. Factorial regression included meteorological, geographic, satellite, and soil variables as predictors. A regression tree algorithm identified the most influential variables, and farmer profiles were grouped using principal component biplots and K-means clustering. Machine learning models proved superior to traditional methods for predicting productivity, offering a strategic tool for agribusiness. Key factors positively associated with yield included mean temperature (around 30°C), relative humidity, longwave and shortwave radiation, high altitude, early sowing, high plant population, elevated soil organic matter, and high cation exchange capacity. Interestingly, yields were higher in soils with magnesium and calcium contents below 13% and 27%, respectively, decreasing beyond those levels. The highest yields (>6 t ha-1) were observed in Rio Grande do Sul, Paraná, São Paulo, and Minas Gerais. Future research should validate these models in low-tech environments and include socio-economic variables.
