Trend & directionArnaud Legoux Moving Average · ALMA
A Gaussian-weighted average with an adjustable offset that trades smoothness against lag.
Works best in trending marketsEngine-computed on a fixed sample series
What it is
The Arnaud Legoux Moving Average, or ALMA, is a modern moving average designed to solve the oldest problem in the field: the trade-off between smoothness and lag. Ordinary averages are either responsive but noisy (like a short EMA) or smooth but slow (like a long SMA); ALMA uses a bell-shaped set of weights that can be slid along the window to get much of the smoothness of a long average with far less of the lag. The question it answers is how to follow the trend closely while still filtering out the jitter that generates false crossovers. It was introduced by Arnaud Legoux and Dimitrios Kouzis-Loukas, and it is prized as a low-lag, low-noise line for use as a trend reference or a crossover signal line.
How it's calculated
ALMA lays a Gaussian (bell curve) set of weights across the lookback window and takes a weighted average of price using those weights. Two parameters shape the curve: the offset, which slides the peak of the bell along the window, and sigma, which controls how sharp or broad the bell is. Pushing the offset toward 1 places the peak over the most recent bars, emphasizing them for a fast, responsive line; pulling it toward 0 places the peak over older bars for a smoother, slower line. A larger sigma makes the bell narrower and the weighting more concentrated, while a smaller sigma spreads the weights more evenly. The window length sets how many bars are included; a common configuration is a window of 9, an offset of 0.85 and a sigma of 6.
Reading it, step by step
You read ALMA like any moving average — by its slope and by price crossing it — but with the understanding that it can be tuned to hug price far more tightly than an SMA of the same length while still filtering more noise than an EMA. A rising ALMA indicates an uptrend and a falling one a downtrend, and its slope changes flag momentum shifts earlier than a comparable lagging average would. Price crossing above the line is a bullish cue and below it a bearish one, and because the line lags less, those crossings arrive sooner. The character of the readings depends heavily on the offset and sigma: a high-offset ALMA gives quick, sometimes jumpy signals, while a lower-offset ALMA gives calmer, later ones. Interpreting it therefore requires knowing how it has been tuned.
Best timeframes and settings
The widely used baseline — window 9, offset 0.85, sigma 6 — produces a responsive yet smooth line that suits swing and intraday trading, and it can be applied from fast scalping charts to daily trend charts. Raising the offset toward 1 or shortening the window makes ALMA faster and better for scalping, at the cost of more whipsaws; lowering the offset or lengthening the window makes it slower and steadier, better as a trend backbone on higher timeframes. Sigma fine-tunes the balance, with larger values concentrating weight on the peak bars for a slightly crisper response. Because ALMA is so tunable, the responsiveness-versus-noise trade-off is set directly through its parameters rather than only through timeframe. The catch is that there is no universally correct setting, so it must be matched to the instrument and timeframe by testing.
When and where to use it
ALMA is a trend tool, so it is most useful in markets that trend and as a smoother, lower-lag replacement for a signal line or a crossover average. It works on any liquid instrument and timeframe, and its tunability lets it adapt to fast crypto charts or slow index charts alike. Use it where the lag of a conventional moving average is causing late entries but a short EMA is too noisy — ALMA occupies that middle ground. In a choppy, non-trending market it still whipsaws like any average, so it does not exempt you from needing a regime filter. Avoid deploying an untuned ALMA blindly, since its behavior swings widely with its parameters; calibrate it to the market first.
Strategies that use it
A crossover strategy uses two ALMAs of different lengths — a fast and a slow — and trades the fast crossing the slow, benefiting from the reduced lag to enter sooner than an equivalent EMA cross. A price-cross trend strategy uses a single ALMA as a dynamic trend line, going long when price closes above a rising ALMA and flat or short when it closes below a falling one, with the line doubling as a trailing reference. A signal-line replacement strategy swaps ALMA in for the smoothing line inside another system — for instance, using an ALMA of an oscillator to get earlier, cleaner trigger crosses. Across these, ALMA's value is that it delivers the earlier signal without the extra noise a short average would add, provided it is tuned sensibly.
Combining it with other indicators
ALMA slots into any setup that uses moving averages, and it pairs well with a trend-strength filter like ADX so that its crossovers are only acted on when a real trend is present. Momentum oscillators such as RSI or MACD confirm the direction of an ALMA crossover and help screen out the whipsaws that any average suffers in a range. Because ALMA is often used as a fast line, a slower conventional average or a higher-timeframe ALMA can serve as the trend filter it trades within. Volatility tools like ATR help set stops around ALMA-based entries, sizing the stop to the market's noise. As with any average, pairing it with support-and-resistance keeps its signals anchored to structure.
Where it fails
ALMA's flexibility is also its trap: its behavior depends so heavily on the offset and sigma you choose that a poorly tuned line can lag badly or whipsaw wildly, and there is no single correct setting to fall back on. Like every moving average it is a lagging, trend-following construct, so in a sideways market it will generate false crossovers no matter how cleverly it is weighted. A high offset that emphasizes recent bars can make it overshoot and reverse on spikes, while a low offset can make it too sluggish to be useful. Traders sometimes over-optimize the parameters to past data and then find the tuning fails going forward. The remedies are to calibrate ALMA to the specific instrument and timeframe, to use a regime filter so it is not trading in chop, and to resist curve-fitting the parameters to history.
A worked example
A swing trader applies an ALMA with window 9, offset 0.85 and sigma 6 to the daily chart of a trending stock, using it as a dynamic trend line. The stock has been climbing, and each pullback has found the ALMA and turned back up, the line lagging price far less than the 20-day SMA the trader used before. Price dips to touch the ALMA at $48 and closes back above it the same day; the trader adds to the long, with a stop just below the line. Because the offset is high, the ALMA had flattened only briefly during the dip rather than rolling over as a slower average would have, so it did not shake the trader out. Price resumes to $54, and the ALMA continues to track just beneath it — demonstrating the tool's selling point: it stayed close enough to give an early re-entry without generating a false exit on the shallow pullback.