KAMA

What is KAMA?

KAMA, Perry Kaufman's adaptive moving average, is an exponential average whose smoothing constant changes bar by bar with how efficiently price is moving. The sensor is the Kaufman Efficiency Ratio: net change over the lookback divided by the sum of absolute bar-to-bar changes. A straight-line move scores near 1; a path that backtracked all the way scores near 0. KAMA maps that score between a fast and a slow EMA constant (commonly 2-period and 30-period equivalents), squares it, and uses the result as the weight on new price.

The squaring matters: middling efficiency gets pushed toward the slow end, so KAMA spends choppy periods nearly flat, ignoring noise, and only accelerates when direction is genuinely persistent. That makes flatness informative in its own right, a built-in no-trend flag. The cost is adaptation lag: the efficiency ratio is computed over a window, so KAMA needs several bars of clean movement before it believes a new trend, and fresh breakouts from quiet ranges are exactly where it starts slow.

How to calculate KAMA

Kaufman's published recipe has three moving parts:

1. 1Efficiency Ratio: ER = absolute net price change over the last N bars divided by the sum of absolute one-bar changes over the same window (N = 10 is the common default), giving a value between 0 and 1.
2. 2Smoothing constant: SC = [ER × (fast − slow) + slow]², where fast and slow are the EMA constants 2/(2+1) and 2/(30+1) in the common parameterization.
3. 3Recursion: KAMA today = KAMA yesterday + SC × (price − KAMA yesterday), the standard EMA update with the adaptive weight.
4. 4Read the output together with its own slope: a sloping KAMA tracks the trend, a flat KAMA declares the window inefficient.

How traders use it

• As a self-filtering trend line: price relative to KAMA plus KAMA's slope, with the flat-line state acting as a stand-aside signal that fixed-length averages cannot provide; it doubles as a crude trend/range classifier.
• In crossover and trailing logic: Kaufman himself suggested acting only when the average turns by more than a small filter (a fraction of the standard deviation of its own recent changes) rather than on every wiggle.
• As a smoothing engine elsewhere: applying KAMA to an oscillator or another indicator's output quiets rangebound noise while keeping trend response, the same adaptivity transplanted.
• As a regime dial via its input: the efficiency ratio itself is worth plotting, and many traders treat low-efficiency stretches as no-trade zones regardless of what the average does.

KAMA vs other adaptive averages