Arnaud Legoux Moving Average (ALMA)

A Gaussian-weighted moving average designed to reduce lag while keeping the line smooth.

What it is

The Arnaud Legoux Moving Average (ALMA) is a moving average developed in 2009 by Arnaud Legoux and Dimitris Kouzis-Loukas. Like any moving average, it plots a single smoothed line over price. What sets ALMA apart is how it weights the data: instead of weighting every bar equally (as a simple moving average does) or weighting recent bars most heavily in a fixed way (as an exponential moving average does), ALMA applies a Gaussian (bell-curve) weighting whose peak can be shifted along the lookback window. The aim is to balance two things that usually trade off against each other in a moving average: smoothness (less reaction to noise) and lag (how far the line trails behind price). ALMA is a descriptive smoothing tool. It summarizes past prices and is not predictive of future prices.

How it works

ALMA takes a lookback window of bars and assigns each bar a weight from a Gaussian (bell-shaped) curve, then sums the weighted prices to produce one value per bar. Three settings control the shape and position of that curve. The "window" (length) sets how many bars are included. "Offset" (typically 0 to 1) slides the peak of the bell curve along the window: an offset near 1 places the peak on the most recent bars, which reduces lag but makes the line more reactive (noisier); an offset near 0 places the peak on older bars, which adds smoothness and lag. "Sigma" controls how wide or narrow the bell is: a smaller sigma concentrates weight tightly around the peak (more responsive, less smooth), while a larger sigma spreads weight out (smoother, more lag). By tuning offset and sigma together, a trader can dial the curve toward responsiveness or smoothness. In Pine Script, ALMA is available as the built-in ta.alma(series, length, offset, sigma).

How traders read it

• Traders often read the slope of the ALMA line as a rough indication of recent trend direction over the chosen window. Slope describes past price behavior and is not a forecast.
• Some traders watch where price sits relative to the line, treating price above as a context of recent strength and below as recent weakness, while keeping in mind the line lags.
• Crossovers between price and ALMA, or between two ALMAs of different lengths, are sometimes used as study points for possible momentum shifts. They are observations about the past, not instructions to act.
• Because offset and sigma alter responsiveness, a high-offset/low-sigma ALMA is typically read as a faster, noisier line and a low-offset/high-sigma ALMA as a slower, smoother one.
• Many traders treat ALMA as one input among several, comparing it against price structure and other tools rather than reading it alone.

Common settings

A common starting point is a length of 9 to 21 with offset around 0.85 and sigma around 6, but there is no universally correct setting. Higher offset and lower sigma make the line faster and noisier; lower offset and higher sigma make it smoother and slower. Settings should be chosen for the timeframe and instrument and treated as a tradeoff, not an optimization target.

Strengths

• Aims to reduce lag for a given level of smoothing compared with simple or exponential moving averages.
• The Gaussian weighting can suppress short-term noise while still tracking the broader move.
• Offset and sigma give granular control over the smoothness-versus-responsiveness tradeoff.
• Available as a built-in (ta.alma) in Pine Script, so it is simple to plot and combine with other tools.

Pitfalls to watch

• ALMA still lags. Reducing lag relative to other moving averages is its design goal, but it cannot eliminate lag, and it never leads price.
• Results are highly sensitive to the offset and sigma settings. The same length can behave very differently, and tuning these on past data can produce a curve that fits history without behaving the same way going forward (curve-fitting).
• In ranging or choppy markets, the line can whipsaw, producing frequent direction changes and crossovers that do not correspond to sustained moves.
• A smoother-looking line is not a more accurate or predictive one. Smoothness and lag trade off, and nothing about the calculation forecasts future prices.
• ALMA describes price only; it carries no information about volume, volatility, or risk. Trading on any indicator carries the risk of loss.

Pine v6 example

//@version=6
indicator("ALMA Example", overlay = true)

length = input.int(21, "Length", minval = 1)
offset = input.float(0.85, "Offset", minval = 0.0, maxval = 1.0)
sigma  = input.float(6.0, "Sigma", minval = 0.1)

almaLine = ta.alma(close, length, offset, sigma)

plot(almaLine, "ALMA", color = color.blue, linewidth = 2)

// Educational only. Not advice, not a signal. Nothing here is predictive; trading carries risk of loss.

Pro tip: If you adjust ALMA's settings, change offset and sigma deliberately and one at a time so you can see what each does, rather than searching for the combination that looks best on past data. A curve that fits history neatly is not evidence it will behave the same way later.

Educational only — not financial advice, not a recommendation to trade. No indicator is predictive; trading involves substantial risk of loss.