Trend indicators

Relative Momentum Index (RMI): Formula, Settings and How It Compares With the RSI

The Relative Momentum Index is the RSI computed on changes over several days instead of one. Here is Roger Altman's definition, the formula, and what momentum settings of 3 and 5 do to the time spent above 70 and below 30 on Bitcoin and the S&P 500 ETF, with three turning points and what followed.

· Founder of PatternsWizard
Checked 7 min read
RMI(20, 5) on Bitcoin, daily, August 30, 2026 to October 28, 2026. Illustration of how the indicator reads, not a backtest.
On this page
  1. How the RMI is calculated
  2. RMI with momentum 3 and 5 against the RSI 14
  3. Three turning points, with the full outcome
  4. Settings
  5. Does it work?
  6. Frequently asked questions

The Relative Momentum Index (RMI) is the RSI computed on price changes over several days instead of one. Roger Altman presented it in Technical Analysis of Stocks & Commodities in February 1993, as a modification of Welles Wilder's 1978 RSI, which, the article's summary notes, “doesn't oscillate evenly”. Where the RSI compares today's close with yesterday's, the RMI compares it with the close a set number of days back, so the line swings further and spends longer at its extremes.

Relative Momentum Index (20, 5) on daily Bitcoin

RMI(20, 5) on Bitcoin, daily, February 21, 2026 to October 28, 2026. Illustration of how the indicator reads, not a backtest.
  • Price, daily candles (hollow up, filled down)
  • RMI(20, 5) (solid line)
RMI(20, 5) on Bitcoin, daily, February 21, 2026 to October 28, 2026. Illustration of how the indicator reads, not a backtest.

How the RMI is calculated

ProRealCode quotes Altman's description: the RSI “is modified by counting up and down days starting from today's close relative to the close Y days ago”, and his example uses a 20-day RMI with Y = 5. The rest is the RSI calculation.

With a momentum of 1, the formula is exactly the RSI: we checked that the two lines match on our data. A momentum of 3 or 5 makes each “up” or “down” value a 3 or 5-day change, which is larger and more persistent than a daily one, so the averages separate more and the line spends longer beyond 70 or 30.

What the RSI and the RMI compare

What the RSI and the RMI compare: a row of closes ending today. A short dashed step compares today's close with the close one bar back, as the RSI does. A long bracket compares it with the close five bars back, as the RMI does with a momentum of 5: that change is larger and keeps its sign longer.
Schematic, not market data.

Each bar's change is counted once, as an up or a down amount, exactly as in the RSI; only the distance of the comparison changes. Because consecutive 5-day changes share four of their five days, they tend to have the same sign for several bars in a row, which is what keeps the line near its extremes.

RMI with momentum 3 and 5 against the RSI 14

We counted, on the same 534 daily bars of each market, the bars spent above 70 and below 30 and the number of entries into each zone. An entry is a bar inside the zone whose previous bar, also within the 534, was outside it; a zone already entered on the first bar is not counted. Bitcoin runs from May 13, 2025, the S&P 500 ETF from September 12, 2024, both to October 28, 2026.

SettingBitcoin, bars above 70Bitcoin, bars below 30Bitcoin, entries above 70 / below 30ETF, bars above 70ETF, bars below 30ETF, entries above 70 / below 30
RSI(14)292210 / 541714 / 4
RMI(14, 3)1037212 / 71542814 / 3
RMI(14, 5)13511111 / 82144318 / 4
RMI(20, 5)117938 / 71843611 / 3

Bars spent above 70 out of 534, by setting

Bars spent above 70 out of 534, by setting. RSI(14): Bitcoin 29, S&P 500 ETF 41; RMI(14, 3): Bitcoin 103, S&P 500 ETF 154; RMI(14, 5): Bitcoin 135, S&P 500 ETF 214; RMI(20, 5): Bitcoin 117, S&P 500 ETF 184. Our computation; the inputs and the method are in the text.
  • Bitcoin
  • S&P 500 ETF
Our computation; the inputs and the method are in the text.Daily closes to October 28, 2026: Bitcoin from May 13, 2025, the S&P 500 ETF from September 12, 2024, dividend-adjusted. Wilder's smoothing for every line.

The number of entries changes little, the time spent in the zones changes a lot: RMI(14, 5) spent 135 of 534 Bitcoin bars above 70, against 29 for the RSI, and 214 bars above 70 on the ETF, about 40% of the period. Read as overbought, that would have meant staying out of a rising ETF for long stretches. Read as a trend filter, the time above 70 is simply the time the trend was up.

Three turning points, with the full outcome

  • Bitcoin low, June 30, 2026 (58,624.71). RMI(20, 5) first went below 30 on June 1 at 71,408.90, 21.81% above the eventual low; the RSI on June 2 at 66,760.83. The RMI left the zone briefly on June 16 at 65,675.01, went back below 30 on June 19, and came back above 30 for good on July 4 at 63,144.01, 7.71% above the low. Twenty days later Bitcoin was up 1.58% and forty days later 0.55%: the exit from oversold did not start a rally until the August breakout.
  • Bitcoin high, October 6, 2025 (124,658.54). RMI(20, 5) entered the zone above 70 on October 5 and left it on October 10 at 112,774.50; the RSI left it on October 7. Forty days after October 10, Bitcoin was 18.82% lower, and three days later it made a low close of 84,739.74 on November 22.
  • S&P 500 ETF low, April 8, 2025 (488.38, dividend-adjusted). RMI(20, 5) first went below 30 on March 6 at 561.68 and the RSI on March 11 at 545.21, before a further fall of 13.05% and 10.42%. The RMI left the zone on March 24, went back below 30 on April 3, and came back above 30 on April 15 at 528.84; the ETF was 12.30% higher forty days later.

Distance from the turning point when RMI(20, 5) entered and left its zone

Distance from the turning point when RMI(20, 5) entered and left its zone. Bitcoin low, June 30, 2026: First entry +21.81%, Final exit +7.71%; Bitcoin high, October 6, 2025: First entry −0.94%, Final exit −9.53%; S&P 500 ETF low, April 8, 2025: First entry +15.01%, Final exit +8.28%. Our computation; the inputs and the method are in the text.
  • First entry
  • Final exit
Our computation; the inputs and the method are in the text.Close on the day of the first entry into the zone and of the final exit from it, compared with the close of the turning point. Below 30 for the lows, above 70 for the high.

At the two lows the line reached its zone weeks before the turn, at the Bitcoin high one day before; in all three cases it left the zone after the turn: it described the move, it did not time it. Three episodes are examples, not a test.

Settings

Two inputs: the length of the averages, 14 as in the RSI or 20 as in Altman's example, and the momentum, the number of days in each change, 5 in his example. A momentum of 1 gives back the RSI. The passages of Altman quoted here do not say which average he used; the author of the ProRealCode version describes an exponential average and a default momentum of 3. The chart uses Wilder's smoothing, as the RSI does, and platforms differ, so check the documentation of yours.

Does it work?

The RMI is a smoother, wider-swinging RSI; nothing published shows it gives better signals. Our bench measures chart patterns detected by rule against random entries, not indicators (see our methodology). For the original, see the RSI; for other ways to read momentum, the momentum indicator and the Trend Intensity Index. A different oscillator with a similar name is the Relative Vigor Index. More tools are listed with the trend indicators, including the MACD. Divergences on the original RSI, and what they did on real data, are covered in RSI divergence trading.

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 Relative Momentum Index?

An RSI built on price changes over several days instead of one, introduced by Roger Altman in 1993. It reads from 0 to 100 like the RSI and swings further toward the extremes.

What is the difference between RMI and RSI?

The RSI compares each close with the previous one; the RMI compares it with the close m days back. With m = 1 they are the same. With m = 5, the line spent several times more bars above 70 and below 30 on our data.

What are the best RMI settings?

No tested best setting is published. Altman's example is a 20-day RMI with a momentum of 5. Before using 70 and 30, check how long the line stays beyond them on your market.

Is there an RMI strategy?

The usual readings are those of the RSI: levels, crosses of 50 and divergences. In three Bitcoin and ETF turning points, the RMI reached its zone before the turn and left it after, so it worked better as a description of the trend than as a timing signal.

Sources

  1. Roger Altman, Relative Momentum Index: Modifying RSI, Technical Analysis of Stocks & Commodities, February 1993
  2. ProRealCode, Relative Momentum Index (RMI), with Altman's description
, 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 trend indicators articles