Trend indicators

ALMA Indicator: Arnaud Legoux Moving Average Formula, the 9, 0.85, 6 Settings and How It Compares

The Arnaud Legoux Moving Average weights the window with a bell curve whose peak the offset slides toward the newest close; here is the formula, what each setting does, and how ALMA tracked daily Bitcoin against simple, exponential and Hull averages of the same length.

· Founder of PatternsWizard
Checked 6 min read
ALMA(9, 0.85, 6) on Bitcoin, daily, August 30, 2026 to October 28, 2026. Illustration of how the indicator reads, not a backtest.
On this page
  1. How ALMA is calculated
  2. What the three settings do
  3. ALMA against the EMA, the SMA and the Hull average
  4. How traders use ALMA
  5. Does it work?
  6. Frequently asked questions

The Arnaud Legoux Moving Average (ALMA) is a moving average whose weights follow a Gaussian bell curve laid across the window, with the peak of the bell shifted toward the most recent bars. Arnaud Legoux and Dimitrios Kouzis-Loukas introduced it in 2009. The idea is a compromise between smoothness and lag: a simple average weights every bar equally and reacts late, while ALMA concentrates its weight near the newest bars and tapers it off on both sides of the peak.

ALMA(9, 0.85, 6) and SMA(9) on daily Bitcoin

ALMA(9, 0.85, 6) on Bitcoin, daily, July 1, 2026 to October 28, 2026. Illustration of how the indicator reads, not a backtest.
  • Price, daily candles (hollow up, filled down)
  • ALMA(9, offset 0.85, sigma 6) (solid line)
  • SMA(9), for comparison (dashed line)
ALMA(9, 0.85, 6) on Bitcoin, daily, July 1, 2026 to October 28, 2026. Illustration of how the indicator reads, not a backtest.

How ALMA is calculated

LuxAlgo gives the steps. Number the bars of the window from 0, the oldest, to n − 1, the newest. The peak of the bell is at m = offset × (n − 1). Its width is s = n / sigma. Each bar i gets the weight exp(−(i − m)² / (2s²)). ALMA is the sum of each close times its weight, divided by the sum of the weights. The defaults, n = 9, offset 0.85 and sigma 6, follow Arnaud Legoux’s original release according to LuxAlgo.

Weight of each close in a 9-bar ALMA, sigma 6

Weight of each close in a 9-bar ALMA, sigma 6. Newest close: Offset 0.85, the default 22.05%, Offset 0.5, centered 0.76%; 1 bar back: Offset 0.85, the default 30.09%, Offset 0.5, centered 3.61%; 2 bars back: Offset 0.85, the default 26.34%, Offset 0.5, centered 10.96%; 3 bars back: Offset 0.85, the default 14.78%, Offset 0.5, centered 21.34%; 4 bars back: Offset 0.85, the default 5.32%, Offset 0.5, centered 26.66%; 5 bars back: Offset 0.85, the default 1.23%, Offset 0.5, centered 21.34%; 6 bars back: Offset 0.85, the default 0.18%, Offset 0.5, centered 10.96%; 7 bars back: Offset 0.85, the default 0.02%, Offset 0.5, centered 3.61%; 8 bars back, the oldest: Offset 0.85, the default 0.00%, Offset 0.5, centered 0.76%. Our computation; the inputs and the method are in the text.
  • Offset 0.85, the default
  • Offset 0.5, centered
Our computation; the inputs and the method are in the text.Share of the total weight, from the newest close to the oldest. At 0.85 the bell peaks one bar back; at 0.5 it peaks on the middle bar and the newest close carries less than 1%.

What the three settings do

SettingDefaultEffect
Window n9Number of bars averaged; longer is smoother and slower
Offset0.85Where the peak sits, from 0 (oldest bar) to 1 (newest); toward 1 the line follows price more closely, at 0.5 the bell is centered and the line is smoothest
Sigma6Width of the bell; a larger sigma narrows it around the peak, a smaller one spreads the weight toward an equal weighting

ALMA against the EMA, the SMA and the Hull average

We computed four 9-bar averages on the same daily Bitcoin closes, July 1 to October 28, 2026, the window of the chart above: ALMA with its defaults, an exponential average, a simple average, and a Hull moving average. To measure lag without real noise, we also ran each average on a made-up series that jumps from 100 to 110 and stays there. The change in slope is the mean absolute change, from one bar to the next, of the bar-to-bar change of the average, divided by the close: the rougher the line, the higher it is. The halfway count is the number of bars after the jump bar until the average passes 105.

Closeness against smoothness, four 9-bar averages

Closeness against smoothness, four 9-bar averages. ALMA (0.85, 6): Mean distance from the close 1.21%, Mean change in slope 0.38%; Exponential: Mean distance from the close 1.73%, Mean change in slope 0.28%; Simple: Mean distance from the close 2.08%, Mean change in slope 0.23%; Hull: Mean distance from the close 1.18%, Mean change in slope 0.55%. Our computation; the inputs and the method are in the text.
  • Mean distance from the close
  • Mean change in slope
Our computation; the inputs and the method are in the text.Daily Bitcoin, July 1 to October 28, 2026. Lower is better for both measures, and no average is lowest on both: the closer a line follows price, the rougher it gets.

The full comparison, with the made-up jump:

9-bar averageMean distance from the closeTimes the close crossed itMean bar-to-bar change in slope, % of priceMade-up jump: bars until halfwayMade-up jump: overshoot
ALMA (0.85, 6)1.21%340.381None
Exponential1.73%250.283None
Simple2.08%240.234None
Hull1.18%400.551Up to 112.67

ALMA stayed much closer to price than the simple and exponential averages of the same length and was crossed by the close more often, the cost of following price closely. The Hull average was slightly closer still, but its slope changed about half as much again from bar to bar, and on the jump it went above 110 before settling. LuxAlgo describes this behavior: the Hull design can overshoot at turns, while ALMA’s off-center bell is a gentler way to reduce lag. ALMA’s line never goes beyond the highest or lowest close of its window, because its weights are all positive.

How traders use ALMA

ALMA is used like any moving average: its slope as the trend, the close crossing it as a change, and two ALMAs of different lengths as a crossover. Because it follows price closely, a 9-bar ALMA gives more crossings than a 9-bar simple average, 34 against 24 in our window; a longer window or a lower offset gives fewer.

Moving averages are covered in more depth in our moving average guide. To go further on averages in general, Steve Burns teaches two New Trader University video courses, Moving Averages 101 and Moving Average Signals (affiliate links).

Steve Burns, founder of New Trader U

Does it work?

We have no long-run measurement of ALMA: our bench tests chart patterns detected by rule against random entries, not indicators (see our methodology). The comparison above measures how closely each average follows price on one market, not whether trading it makes money.

For other low-lag averages, see the Hull moving average and the volume-weighted moving average, or the full list of trend indicators.

Every pattern's results on one sheet

The only pattern cheat sheet with tested stats: 152 patterns, what the textbook says and what really happened on 6 markets. Free PDF, sent to your inbox.

The PDF by email, plus our newsletter: new tests and results that change. Unsubscribe anytime. Privacy.

Frequently asked questions

What is the ALMA indicator?

The Arnaud Legoux Moving Average, a moving average whose weights follow a Gaussian bell curve shifted toward the newest bars. It follows price more closely than simple and exponential averages of the same length, at the cost of a slightly rougher line.

What are the best ALMA settings?

The defaults are 9 bars, an offset of 0.85 and a sigma of 6, from the original release. A longer window or a lower offset smooths more. No setting has been shown to be the best for all markets.

Is ALMA better than the EMA?

On our Bitcoin window, a 9-bar ALMA stayed closer to price than a 9-bar EMA, and it was crossed by the close more often. Closer is not the same as more profitable, and we know of no test showing ALMA trades better.

What is the difference between ALMA and the Hull moving average?

Both reduce lag. Hull combines weighted averages and can overshoot a sharp move; ALMA is a single average with positive weights, so it cannot go beyond the closes of its window.

Does ALMA repaint?

No. Each value uses only the closes of its window, up to the current bar, so a past value does not change when new bars arrive.

Sources

  1. LuxAlgo, Arnaud Legoux Moving Average (ALMA)
  2. Kolkova, A. (2018), Indicators of Technical Analysis on the Basis of Moving Averages as Prognostic Methods in the Food Industry, Journal of Competitiveness
, Founder of PatternsWizard

Passionate about finance, quantitative trading and data analysis. Amaury built the PatternsWizard test bench, which detects each pattern on full price history and measures what happened next against random entries.

PatternsWizard tests chart and candlestick patterns on full market history, against random entries. How we test patterns.

179 patterns testedRandom-walk self-test passed

All trend indicators articles