This document discusses two types of complexity that can affect trend detection in time series data: long range dependence and heavy tails. Long range dependence, if present in a system, implies the presence of low frequency "slow" fluctuations that can complicate trend detection. Heavy tails in a probability distribution are a source of "wild" fluctuations due to more frequent extreme events. The document reviews several examples of long range dependence and heavy tails observed in real-world datasets like financial data and space weather data. Statistical models like linear fractional stable motion (LFSM) and autoregressive fractionally integrated moving average (ARFIMA) processes are discussed for modeling systems with both properties. Better statistical inference methods are also needed to distinguish true