Market data is sometimes referred to as "colored noise" (in contrast to "white noise") or 1/f noise. The underlying processes that drive market movement are mixed with random data. Market models frequently smooth the market time series in an attempt to reduce the amount of noise in the data and reveal the underlying movements. Smoothing the market time series is particularly important for software models that take market data as inputs.

Many of the smoothing filters introduce lag. Change in the filtered occurs later in time than change in the time series. This can create problems for trading models, since it is possible that a trading model that uses smoothed data will react too late, missing profit opportunities. Smoothing functions that operate on large data windows usually produce smoother data that reflects only the overall time series movement, which is desirable for trading models. However, these models can have significant lag, which means that by the time the model can act on the change, the event may be too far in the past. Smoothing versus lag are the two design characteristics that are traded off with this type of smoothing model.