We investigate how EEG model performance scales with participant count and total sample size, and assess the impact of data augmentations and pre-training across different data regimes.