A trading model whose internal logic is opaque or complex (often machine-learning models). Black-box algorithms (such as deep neural nets or large ensembles) generate signals from data using complex nonlinear transformations, but they do not provide easy explanations for why a decision was made. In Forex cBots, a black-box signal module might outperform simpler rules, but it cannot be easily audited or interpreted by the user.