Moving-average windows before crossovers and MACD

A declared lookback, not a freehand line

A moving average is a mechanical summary of an ordered price series over a stated lookback. It is a fixed arithmetic construction, so it includes every observation inside its window instead of omitting inconvenient prints the way a discretionary visual overlay can.

The lookback is a filter window. Prices inside it affect the current value and prices older than the window are excluded, so length is a choice about which history is allowed to speak.

How a simple average moves

A simple average weights each included price equally. Each new bar it adds the current price and drops the oldest, so the window has a hard leading and trailing edge. Comparing today’s price with the departing price indicates whether the average itself will rise or fall.

For a continuous time-ordered series, today’s simple average represents conditions near the middle of the included group, about half the window earlier, not the latest print.

How an exponential average remembers

An exponential average updates by mixing a fraction called alpha of the prior average with the complementary fraction of the current price. Those two weights sum to one. The construction retains a decaying imprint of the entire past series rather than ejecting a single old print.

Exponential smoothing strength and simple-average length can be paired by L = 2/alpha or L = (2/alpha) - 1, which is why the two constructions can track similarly.

Choosing which history is allowed to speak

Markets are described as showing only episodic, not persistent, periodicity. A construction heuristic is to relate average length to that typical period, commonly about half of it. That episodic periodicity is useful as a length hint even though no market is treated as permanently cyclic.

Window length can be judged visually from lag by seeking a line that skims the highs of declines and the lows of advances. Lengthening the window sits farther from swings and shortening it cuts closer.

Trial starting lengths given for that visual fit are 21 days for stocks, 18 days for agriculturals, 10 days for financials, and 20 days for currencies, and they are adjustable. Some series may need different windows for advances and declines.

Crossovers and MACD as later overlays

Crossing one average with another is a later-stage signal built on the same windowed construction rather than on a different data idea. MACD is a further average-based overlay that compares smoothed averages in order to make the same baseline reading more timely.

Both are framed as elaborations of the same filter meant to make the reading more timely rather than as independent replacements.