The Median-Average Adaptive Filter was introduced by John F. Ehlers in the Stocks & Commodities article “The Secret Behind The Filter – What’s The Difference”. It addresses the difference between two types of noise filters, namely the average and the median.
Indicator Description
The average will attempt to eliminate noise by summing to its own average value, whereas the Median Average Adaptive Filter ( MAAF ) simply ignores outliers (noise) altogether. Accordingly, a big spike within a median data set will have no effect on it’s output. For example, if we have a dataset of 10 consecutive 1s, both the average and median is 1. However, if the next value in the dataset is 10, the new average is 1.9 (9×1 + 10 / 10 = 1.9), whereas the median remains unchanged at 1. This is because the median filter ranks all samples within the dataset, using the middle one as its filter output.
You may start with a relatively large lookback period. One should use an odd number to ensure that the MAAF indicator will be “centered”. The filter will respond to larger moves but does not move significantly during consolidations, thereby avoiding noise signals sideways markets.
For further details on the MAAF indicator, please refer to the Stocks & Commodities V. 23:3 (28-29): What’s The Difference by John F. Ehlers.
Other Library indicators
Other indicators from this category include the Adaptive Laguerre Filter, the Butterworth Filter, the Coral Filter, the Distant Coefficient Filter, the Gaussian Filter, the Laguerre Filter and the Supersmoother Filter. You may also review our indicators from the moving averages or the trend analysis categories, such as the Rainbow Filter, Directional Movement Index, Sine WMA, Range Weighted Moving Average or the Efficiency Ratio. The Indicator Spotlight newsletter furthermore discussed the Efficiency Ratio here.
The Median Average Adaptive Filter, i.e. MAAF indicator is available for NinjaTrader 8.