<p>To test hypotheses about glacial dynamics, the Mid-Brunhes event, and the stage 11 paradox, we evaluate the ability of a statistical model to simulate climate during the previous ~800 000 years. Throughout this period, the model simulates the timing and magnitude of glacial cycles, including the saw-tooth pattern in which ice accumulates gradually and ablates rapidly, without nonlinearities or threshold effects. This suggests that nonlinearities and/or threshold effects do not play a critical role in glacial cycles. Furthermore, model accuracy throughout the previous ~800 000 years suggest that changes in glacial cycles associated with the Mid-Brunhes event, which occurs near the division between the out-of-sample period and the in-sample period, are not caused by changes in the dynamics of the climate system. Conversely, poor model performance during MIS stage 11 and Termination V is consistent with arguments that the <q>stage 11 paradox</q> represents a mismatch between orbital geometry and climate. Statistical orderings of simulation errors indicate that periods of reduced accuracy start with significant reductions in the model's ability to simulate carbon dioxide, non-sea-salt sodium, and non-sea-salt calcium. Their importance suggests that the stage 11 paradox is generated by changes in atmospheric and/or oceanic circulation that affect ocean ventilation of carbon dioxide.</p>