How a price-hugging smoother is assembled from ordinary averages

Fixed windows and ordinary averages

A simple moving average is a fixed-length smoother of ordered prices. It needs a full window of past points and cannot change its length as the series speeds up or slows down.

A moving average is a rolling smoother of ordered prices or breadth that flattens a series over a chosen lookback. Two named problems of ordinary averages are inappropriate application and overuse, including being outrun by raw data when they are treated as trading systems.

Exponential smoothing as a two-input update

Early moving-average practice treated fitting lines and graduation as ways to interpolate ordered series, later joined by exponential inventory-control smoothers.

Exponential smoothing updates a prior trend with the latest observation using the constant 2/(n+1). That smoothing constant is the weight that turns an n-day exponential average into a percent-of-trend update, so a 19-day average is a 10 percent trend.

Assembling the McGinley dynamic

An adaptive moving average is a smoother whose update speed changes with the current price-to-line ratio instead of staying a fixed length.

The McGinley update is prior dynamic plus the index-minus-dynamic gap divided by N times the fourth power of the index-to-dynamic ratio. The n-parameter is the user-chosen speed constant that keeps the McGinley update inside a percentage band.

The fourth-power ratio is the adjustment that lets the line accelerate, especially on downward moves, by enlarging the gap between the line and current data.

Built to hug prices, not to signal

The construction goal is a smoother that hugs prices, reduces whipsaws, and follows both fast and slow markets instead of acting as a signal generator. A whipsaw is a false turn created when a smoother lags or overshoots raw prices.