Kaufman Adaptive Moving Average (KAMA) Guide

KAMA's core innovation is a single measurement called the Efficiency Ratio (ER). Think of it like a GPS comparing your straight-line distance to your actual driving distance — the more winding the road, the less efficient the journey. In market terms, ER divides the net directional price change over a period by the total path length of all individual price moves during that same period.

With the default 10-period setting, KAMA looks back 10 bars. If price moved 50 pips net over those 10 bars but traveled a total of 200 pips back and forth, the ER is 0.25 — low efficiency, meaning the market is noisy. If price moved 50 pips net and only traveled 55 pips total, the ER is 0.91 — high efficiency, meaning a clean trend is in place.

That ER value then feeds into a Smoothing Constant (SC), calculated using the fast period (default: 2) and slow period (default: 30). When ER is high, SC pulls toward the fast EMA equivalent — a 2-period EMA reacts in roughly 2 bars. When ER is low, SC drops toward the slow EMA equivalent — a 30-period EMA barely moves. The final KAMA value is: KAMA(today) = KAMA(yesterday) + SC² × (Price − KAMA(yesterday)).

The squaring of SC is deliberate. It amplifies the difference between trending and ranging states, making KAMA dramatically more responsive in trends and dramatically flatter in noise — unlike a standard EMA, which applies the same multiplier regardless of market conditions. This non-linear behavior is what separates KAMA from every fixed-period moving average.

Counterintuitive as it sounds, the most reliable KAMA signals often come not from price crossing the line, but from the slope of KAMA itself changing direction. A flat KAMA line is telling you something explicit: the market is going nowhere worth trading.

Buy signals appear when KAMA turns upward after a flat or declining period and price is trading above the KAMA line. Compared to a standard 20-period EMA crossover, this approach generates fewer signals — but each signal carries more weight because the indicator has already filtered out the surrounding noise before committing to a direction.

Sell signals are the mirror image: KAMA turns downward after a flat or rising period with price below the line. The slope change is the trigger, not merely the price-to-KAMA relationship.

Divergence setups add a second layer of confirmation. When price makes a higher high but KAMA makes a lower high — or fails to extend its own high — the efficiency of the trend is deteriorating. This divergence between raw price action and KAMA's adaptive reading often precedes reversals by 3 to 8 bars on H4, giving enough lead time to tighten stops or reduce position size.

One practical filter: measure the distance between price and KAMA. During strong trends on D1, price typically stays 0.3% to 1.2% above or below KAMA. When price extends more than 2% away from KAMA on a D1 chart, mean-reversion back toward the line becomes statistically more probable than continuation — a useful context for scaling out of existing positions rather than entering new ones.

The default parameters — period 10, fast period 2, slow period 30 — were designed with daily charts in mind, and they perform best on D1 and H4 where intraday noise is naturally filtered by the timeframe itself. On those timeframes, the 10-period lookback spans 2 trading weeks on D1 and roughly 40 hours on H4, providing enough data for the ER calculation to distinguish genuine trends from consolidation.

On H1, the default settings can produce excessive flat periods during the Asian session when volatility compresses. Reducing the period to 8 and the slow period to 20 makes KAMA slightly more responsive without sacrificing its adaptive quality. Compared to using a 20-period EMA on H1 — which reacts to every 30-minute noise spike — this adjusted KAMA still filters roughly 40% more false crossovers in typical forex pairs.

For D1 swing traders, some practitioners extend the period to 14 and the slow period to 50. This configuration keeps KAMA nearly flat during multi-week consolidations that trap EMA-based systems, then accelerates sharply when a genuine breakout registers high ER values. The tradeoff is a slightly later entry — typically 1 to 2 days after a breakout begins — in exchange for a dramatically cleaner signal.

H4 with the default settings occupies the sweet spot for most trend-following strategies. The 10-period ER calculation spans roughly 40 hours of price action, long enough to identify multi-day trends while short enough to respond to trend changes within the same trading week. Pairs like EUR/USD and GBP/USD show particularly clean KAMA behavior on H4 because their volatility patterns are well-distributed across sessions.

A KAMA-based system works best when paired with a volatility or momentum filter. The simplest combination is KAMA slope direction plus ATR (Average True Range) threshold: only take KAMA signals when the 14-period ATR is above its own 20-period average, confirming that volatility supports a trending environment. Whereas MACD crossovers fire in both trending and ranging markets indiscriminately, this KAMA-ATR combination produces signals almost exclusively during genuine expansions.

Entry execution follows a two-step process. First, identify KAMA slope turning positive (for longs) or negative (for shorts). Second, wait for the first bar to close beyond KAMA in the direction of the slope — this avoids entering on the exact bar of the slope change, which occasionally whipsaws back. This one-bar confirmation reduces win rate slightly but improves average reward-to-risk from roughly 1.4:1 to approximately 1.9:1 in backtests on EUR/USD D1 from 2015 to 2023.

Stop placement deserves specific attention. Because KAMA flattens in ranging markets, it creates a natural zone of support/resistance. Placing a stop 1.5× ATR beyond the KAMA line at entry time puts the stop outside the indicator's own noise band — not an arbitrary pip count. This is meaningfully different from fixed-pip stops, which ignore the market's current volatility state.

Pulsar Terminal's built-in SL/TP tools make this practical: you can set stop-loss levels based directly on KAMA's position on the chart with a single click, then attach a trailing stop that adjusts as KAMA extends through the trade. Position sizing, too, should scale with the KAMA-to-price distance — wider distance means wider stop, which means smaller size to keep risk constant at, for example, 1% of account equity per trade.

KAMA's adaptive mechanism gives it a structural advantage over fixed-period averages in one specific scenario: transitional markets — environments that shift between trending and ranging within the same chart. A 50-period EMA stays sluggish during a breakout; a 10-period EMA whipsaws during consolidation. KAMA handles both states in a single line.

Unlike the Arnaud Legoux Moving Average (ALMA), which reduces lag through Gaussian distribution weighting but remains fixed in its responsiveness, KAMA actively changes its smoothing coefficient based on current market behavior. ALMA will produce the same lag reduction regardless of whether the market is trending or flat; KAMA will produce near-zero lag in a trend and maximum smoothing in a range.

Compared to the Hull Moving Average (HMA), which prioritizes minimal lag above all else, KAMA is more conservative. HMA on a 14-period setting will react to a 3-bar price move as though it were a trend; KAMA will wait for the ER to confirm directional efficiency before moving. In strong, sustained trends, HMA enters earlier. In choppy markets, KAMA avoids losses that HMA accumulates.

The genuine weakness of KAMA appears in fast-reversing markets — sharp V-shaped recoveries after flash crashes, for example. Because KAMA requires several bars of high ER to accelerate, it can miss the first 30% to 40% of a rapid reversal move. Traders who prioritize catching every reversal will find KAMA frustrating. Those who prioritize riding confirmed trends with minimal drawdown will find it one of the most reliable tools available.

The tradeoff is explicit: KAMA sacrifices early entry for confirmation quality. That exchange is worth making in most systematic trend-following contexts, particularly on H4 and D1 where the cost of false entries — in both capital and psychological terms — compounds quickly over a trading year.