Indicators

GMMA Indicator: How the Guppy Multiple Moving Average Works, With Tests on Four Markets

The Guppy Multiple Moving Average built from its twelve averages, the three states of the two groups, a real Bitcoin example and how GMMA rules did against holding on stocks, gold, Bitcoin and EUR/USD.

· Founder of PatternsWizard
Checked 9 min read
GMMA, EMA 3 to 15 and 30 to 60 on Bitcoin, daily, August 30, 2026 to October 28, 2026. Illustration of how the indicator reads, not a backtest.
On this page
  1. What the GMMA is
  2. How to read the two groups
  3. A real example on Bitcoin
  4. Testing GMMA rules on four markets, up to 33 years of data
  5. Does it work?
  6. Frequently asked questions

The Guppy Multiple Moving Average (GMMA) puts twelve exponential moving averages on the price chart, in two groups of six: a short group from 3 to 15 days and a long group from 30 to 60 days. Instead of a single crossover, it shows how the two groups sit relative to each other: apart and in order in a clean trend, tangled when the market has no direction. This page covers the construction, how to read the two groups, and how simple rules built on them did on four daily markets, compared with holding.

GMMA on daily Bitcoin

GMMA, EMA 3 to 15 and 30 to 60 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)
  • Short group, EMA 3, 5, 8, 10, 12, 15 (solid line)
  • Long group, EMA 30, 35, 40, 45, 50, 60 (solid line, second color)
GMMA, EMA 3 to 15 and 30 to 60 on Bitcoin, daily, July 1, 2026 to October 28, 2026. Illustration of how the indicator reads, not a backtest.

What the GMMA is

Optuma credits the GMMA to Daryl Guppy, who introduced it in his book Trading Tactics. Optuma describes the short group as a proxy for short-term traders and speculators, and the long group as a proxy for long-term investors. The averages are ordinary exponential moving averages of the close; the information is in the twelve lines together, not in any one of them.

How to read the two groups

Optuma puts the main idea in one line: the crossover of the two groups is not as important as the relationship between them. Three states matter:

  • Groups apart, short group above: every short average is above every long average. The trend is up and traders and investors agree.
  • Groups apart, short group below: the mirror image, a downtrend.
  • Groups tangled: some short averages are above some long ones and some below. There is no clean trend.

The three states of the GMMA groups

The three states of the GMMA groups: three panels, each with a bundle of six solid lines for the short group and six dashed lines for the long group. Left: both bundles rise, the short group well above the long one, an uptrend. Middle: the short group weaves through a flat long group, so the groups cross and there is no clean trend. Right: both bundles fall, the short group below the long one, a downtrend.
Schematic, not market data.The relationship between the groups, not one crossover, is what the GMMA reads.

The spacing inside each group adds detail. Optuma reads both groups narrowing as a sign the trend is weakening; LuxAlgo reads wide, orderly separation inside the long group as investor commitment behind the trend. LuxAlgo describes a full crossover of the two groups as a slow structural signal that arrives late, which is the price of using twelve averages up to 60 days long.

A real example on Bitcoin

The GMMA chart under the introduction covers 120 daily bars of Bitcoin from July 1, 2026. It starts in a downtrend: short group below, long group in falling order, the 60-day average on top and the 30-day one at the bottom, the highest 5.47% above the lowest. Through July and August the chart alternated between the short group below and tangled, while the long group compressed until its six lines were within 0.47% of each other on August 20.

How far apart the six long averages were

How far apart the six long averages were. July 1: short group below: 5.47%; July 15: tangled: 3.49%; July 29: tangled: 1.83%; August 12: short group below: 1.21%; August 20: short group above: 0.47%; September 2: short group above: 4.14%; September 16: short group above: 4.07%; September 30: short group above: 4.85%; October 14: tangled: 3.76%; October 28: tangled: 2.05%. Our computation; the inputs and the method are in the text.
Our computation; the inputs and the method are in the text.Daily Bitcoin, 2026, every second Wednesday plus August 20: the highest of the 30 to 60 day averages, in percent above the lowest, with the state of the two groups that day. The long group was at its tightest on August 20, the day the short group cleared it.

On August 20, with the close at 73,025.15, the whole short group was above the whole long group for the first time in the window. A rule that buys at the next open when the groups separate upward and sells when they separate downward entered on August 21 and was still in the trade on October 28, up 11.39%. Along the way the chart turned tangled on October 14, separated upward again on October 20 and was tangled once more from October 23; on October 28 the long group was in rising order, the 30-day average on top and the 60-day one at the bottom, the highest 2.05% above the lowest. In the same 120 bars, the pair one would trade as a simple crossover between the two groups, the 3-day and 30-day averages, crossed 12 times.

Testing GMMA rules on four markets, up to 33 years of data

We tested three long-only rules on the daily history of four markets (EUR/USD to June 30, 2023, see below the table), each signal taken at the close and executed at the next open, with no costs, compared with buying and holding over the same dates:

  • Groups separate: buy when every short average is above every long one, sell when every short average is below every long one. Tangled periods change nothing.
  • EMA 3 over EMA 30: a plain moving average crossover using the fastest line of the short group and the fastest line of the long group.
  • EMA 15 over EMA 30: the slowest short line against the fastest long line.
Market, fromRuleTradesWinning tradesTotal returnMax drawdown
S&P 500 ETF, October 18, 1993Groups separate6149.2%+1,153.5%−35.3%
EMA 3 over EMA 3025337.5%+405.5%−36.9%
EMA 15 over EMA 3011646.6%+731.8%−32.0%
Buy and hold+2,846.1%−55.2%
Gold ETF, August 9, 2005Groups separate4841.7%+322.7%−35.1%
EMA 3 over EMA 3016633.7%+339.7%−32.3%
EMA 15 over EMA 307642.1%+317.9%−36.4%
Buy and hold+763.2%−45.6%
Bitcoin, February 14, 2018Groups separate2343.5%+1,807.0%−47.1%
EMA 3 over EMA 308330.1%+1,875.8%−50.9%
EMA 15 over EMA 304436.4%+1,136.0%−57.6%
Buy and hold+853.0%−76.6%
EUR/USD, September 11, 2012Groups separate2725.9%−10.1%−20.7%
EMA 3 over EMA 309026.7%−4.6%−17.1%
EMA 15 over EMA 304526.7%−11.5%−20.7%
Buy and hold−14.5%−31.2%

Start dates are the first day each rule could be computed. The S&P 500 ETF series is adjusted for dividends. The EUR/USD test ends on June 30, 2023, the last date before gaps of 10 to 101 days in our daily EUR/USD series, which would put false crossovers into the averages; the other three markets run to October 28, 2026.

Total return of the group rule and of holding

Total return of the group rule and of holding. S&P 500 ETF, from October 18, 1993: Groups separate +1,153.5%, Buy and hold +2,846.1%; Gold ETF, from August 9, 2005: Groups separate +322.7%, Buy and hold +763.2%; Bitcoin, from February 14, 2018: Groups separate +1,807.0%, Buy and hold +853.0%; EUR/USD, from September 11, 2012: Groups separate −10.1%, Buy and hold −14.5%. Our computation; the inputs and the method are in the text.
  • Groups separate
  • Buy and hold
Our computation; the inputs and the method are in the text.Long only, signal at the close, trade at the next open, no costs, to October 28, 2026 (June 30, 2023 for EUR/USD). The open trades are counted at the last close.

What the numbers show:

  • The group rule traded far less than the crossover. On the S&P 500 ETF, 61 trades against 253, with 12 trades ending within 2% of their entry price against 141. That is the filtering effect one would expect from twelve averages read together.
  • It did not beat holding on the two assets that rose steadily. On the S&P 500 ETF the group rule returned +1,153.5% against +2,846.1% for holding, while invested about three days in four; on gold, +322.7% against +763.2%. It did cut the worst drawdown.
  • On Bitcoin it beat holding, +1,807.0% against +853.0%, with a worst drawdown of −47.1% against −76.6%. The plain 3/30 crossover did slightly better still.
  • On EUR/USD, a market with no long trend over the period, every rule lost. Holding lost 14.5%.

Worst drawdown of the group rule and of holding

Worst drawdown of the group rule and of holding. S&P 500 ETF: Groups separate −35.3%, Buy and hold −55.2%; Gold ETF: Groups separate −35.1%, Buy and hold −45.6%; Bitcoin: Groups separate −47.1%, Buy and hold −76.6%; EUR/USD: Groups separate −20.7%, Buy and hold −31.2%. Our computation; the inputs and the method are in the text.
  • Groups separate
  • Buy and hold
Our computation; the inputs and the method are in the text.Largest fall from a peak of the account, marked at each daily close, over the same dates as the table. The group rule cut the worst fall on all four markets, at the cost of a lower return on the two that rose for decades.

The share of days in each state is a measure of how often the GMMA has anything to say. The groups were tangled on about a fifth to a quarter of days on all four markets (20.0% to 24.0%); the chart gives the rest.

Share of days with the groups apart, by side

Share of days with the groups apart, by side. S&P 500 ETF: Short group above 60.6%, Short group below 19.3%; Gold ETF: Short group above 47.6%, Short group below 28.6%; Bitcoin: Short group above 40.3%, Short group below 36.1%; EUR/USD: Short group above 34.8%, Short group below 41.2%. Our computation; the inputs and the method are in the text.
  • Short group above
  • Short group below
Our computation; the inputs and the method are in the text.Daily closes over the same dates as the table. The remaining days, tangled, are 20.0% on the S&P 500 ETF, 23.8% on gold, 23.7% on Bitcoin and 24.0% on EUR/USD.

On October 28, 2026 the group rule was in a trade on two of the three markets tested to that date: on the S&P 500 ETF since April 16, 2026 (up 10.72%) and on Bitcoin since August 21 (up 11.39%). The 15/30 rule was also long the S&P 500 ETF, since August 3, 2026 (up 3.31%); no other rule had a trade open. On EUR/USD all three rules were in a trade on June 30, 2023, the last day of its test. Open trades are not in the trade counts of the table; the total returns include them at the last close of each test.

Does it work?

As a trend filter, the GMMA rule behaved in our tests like a slow moving average system: fewer trades than a fast crossover, smaller drawdowns than holding, and in our tests no gain over holding on the markets that trended up for decades. It beat holding on Bitcoin and did nothing on a range-bound currency pair. Our bench measures chart patterns detected by rule against random entries, not indicators (see our methodology); the channel pattern page is an example of a trend structure measured that way.

For related tools, see the moving average, the moving average crossover, rainbow charts and the page on trend trading, or the full list of 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 GMMA indicator?

The Guppy Multiple Moving Average, credited to Daryl Guppy: twelve exponential moving averages in a short group (3 to 15 days) and a long group (30 to 60 days), read by how the two groups sit relative to each other.

What are the GMMA settings?

Short group 3, 5, 8, 10, 12 and 15; long group 30, 35, 40, 45, 50 and 60, all exponential averages of the close, as listed by Optuma.

What is the GMMA buy signal?

The usual reading is the short group crossing and separating above the long group, ideally with the long group widening. LuxAlgo calls the full crossover a slow signal that arrives late.

Is the GMMA better than a moving average crossover?

It traded far less in our tests: 61 trades against 253 for a 3/30 crossover on the S&P 500 ETF since 1993, with fewer flat trades. Returns were higher on that market, similar on gold and Bitcoin, and neither beat holding on stocks or gold.

Does the GMMA work in ranging markets?

Not in our tests: on EUR/USD from 2012 to mid-2023 the group rule lost 10.1% over 27 closed trades, with 25.9% of them winning.

Sources

  1. Optuma, Guppy Multiple Moving Average
  2. LuxAlgo, Guppy GMMA
, 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

Related guides

All indicators articles