Gap Between Real World and Abstract Models Forecasting

What is the gap between the complexity of the real world and the simplicity of many abstract models used in forecasting?