Home / Wiki / K / Kaufman's Adaptive Moving Average (KAMA)

Kaufman's Adaptive Moving Average (KAMA)

Trading and SpeculationUpdated 16 Nov 2026

Kaufman's Adaptive Moving Average (KAMA) is a moving average that changes its own sensitivity according to market conditions. When prices move steadily in one direction it follows them closely, and when they swing back and forth without progress it flattens out and largely ignores the swings. It was developed by the American trading-systems analyst Perry Kaufman and set out in detail in his writing in the 1990s.

The indicator addresses a familiar trade-off in technical analysis. A fast average reacts quickly to a new trend but produces many false signals in a choppy market. A slow average filters out the chop but reacts late. An adaptive average tries to be fast or slow as conditions require. KAMA is plotted on the price chart like any other average and is offered in most charting packages, often under the plain name adaptive moving average.

01

The efficiency ratio

The adjustment is driven by a measure Kaufman called the efficiency ratio. It compares the net price change over a period with the total distance prices travelled to get there:

ER=∣Pt−Pt−n∣∑i=0n−1∣Pt−i−Pt−i−1∣ER = \frac{\left| P_t - P_{t-n} \right|}{\sum_{i=0}^{n-1} \left| P_{t-i} - P_{t-i-1} \right|}
  • PtP_t is the latest closing price.
  • nn is the look-back period, commonly 10.
  • The numerator is the net change over nn periods, and the denominator is the sum of the absolute period-to-period changes.

The ratio lies between 0 and 1. If prices rose every day, the net change equals the total path and the ratio is 1. If prices moved a great deal but ended where they started, it is 0. A high ratio indicates a clean trend and a low ratio indicates noise. The same idea is sometimes described as a signal-to-noise measure, since it sets the useful part of the price movement against the wasted motion around it.

02

Calculation

The efficiency ratio is converted into a smoothing constant that lies between those of a fast and a slow exponential moving average, by default 2 periods and 30 periods:

SC=[ER×(f−s)+s]2SC = \left[ ER \times (f - s) + s \right]^2

where f=2/(2+1)=0.6667f = 2 / (2 + 1) = 0.6667 and s=2/(30+1)=0.0645s = 2 / (30 + 1) = 0.0645. Squaring the result makes the average slow down sharply when the market is inefficient. The average is then updated:

KAMAt=KAMAt−1+SC×(Pt−KAMAt−1)KAMA_t = KAMA_{t-1} + SC \times (P_t - KAMA_{t-1})

The first value is usually seeded with a simple moving average or with the price itself.

03

Worked example

Suppose that over 10 days the price has risen by a net 5 points, while the daily changes, ignoring sign, add up to 10 points. The efficiency ratio is 5/10=0.55 / 10 = 0.5. The smoothing constant is (0.5×(0.6667−0.0645)+0.0645)2=(0.3656)2=0.1337(0.5 \times (0.6667 - 0.0645) + 0.0645)^2 = (0.3656)^2 = 0.1337. If yesterday's KAMA was 100 and today's close is 104, the new value is 100+0.1337×(104−100)=100.53100 + 0.1337 \times (104 - 100) = 100.53.

Had the ratio been 1, the constant would have been 0.66672=0.44440.6667^2 = 0.4444 and the average would have moved to 101.78. Had it been 0, the constant would have been 0.06452=0.00420.0645^2 = 0.0042 and the average would have barely moved, to 100.02.

04

Use in trading

KAMA is read in the same ways as other averages. Its slope indicates the direction of the trend, and a flat line suggests a market in consolidation. Price crossing above or below the line is taken as a possible change of direction. Two KAMA lines of different settings can be combined in a moving average crossover, and the line is also used as a trailing reference for a stop-loss order. Because the line hugs prices in a trend and goes flat in a range, some traders filter signals by requiring it to move by more than a set amount before acting. The indicator is common in trend following systems, and the efficiency ratio is used on its own as a gauge of trend quality.

05

Limitations

KAMA is still built from past prices and belongs with the lagging indicators. It cannot anticipate a reversal, and after a long quiet period it may respond slowly to the first days of a new move, because the efficiency ratio needs several periods to rise. A sudden jump in volatility with no lasting direction can also pull the line briefly the wrong way.

The indicator has three parameters, which gives room to tune it to historical data. Results from backtesting that depend on a narrow choice of settings may reflect overfitting and fail on new data. Comparisons with ordinary averages show that adaptive methods reduce false signals in some markets and periods and not in others, and no setting removes losing trades.

Last updated 16 Nov 2026