Direct Answer
The Relative Momentum Index (RMI) is a variation on RSI developed by Roger Altman that substitutes momentum, the price change over several periods back, commonly five, for the single-period price change RSI normally uses, then applies RSI's same up/down averaging and RS/RSI-style smoothing formula to that momentum series. The result is a 0-100 momentum oscillator that is typically smoother and slower to react than standard RSI.
Key Takeaways
- RMI is RSI with one change: it compares each close to the close n periods earlier instead of just one period earlier, with a 5-period momentum look-back commonly cited as the default.
- After that substitution, RMI runs the same up/down averaging and RS/RSI-style smoothing formula as standard RSI, producing a line on the same 0-100 scale.
- Because it's built from a wider momentum window, RMI is typically smoother and slower to react than single-period RSI, fewer whipsaws, but later signals.
- Since RMI shares RSI's 0-100 scale, some traders apply the same 70/30 overbought/oversold zones by analogy, though this isn't a fixed rule specific to RMI, and a slower line tends to produce fewer, later crossings.
- Like RSI, RMI can stay pinned near an extreme through a strong trend, an overbought or oversold reading describes momentum conditions, not a guaranteed reversal.
What Is the Relative Momentum Index?
The Relative Momentum Index is a momentum oscillator developed by Roger Altman as a direct variation on Welles Wilder's Relative Strength Index (RSI). Standard RSI is built from the daily price change, each day's close compared with the prior day's close. RMI keeps RSI's overall structure but swaps out that single-period comparison for a momentum value: each day's close compared with the close several periods earlier, with a 5-period look-back commonly cited as the default.
Once that momentum series exists, RMI treats it exactly the way RSI treats daily price changes, separating the up moves from the down moves, averaging each side, and running the same RS/RSI-style smoothing formula to arrive at a final reading on the familiar 0-100 scale. The practical effect of comparing prices further apart is a line that reacts less to single-bar noise and more to a sustained multi-period push in one direction, which is why it's typically described as smoother and slower than standard RSI.
Key takeaways: RMI substitutes a multi-period momentum value (commonly a 5-period look-back) for RSI's single-period price change, then applies RSI's same averaging and smoothing formula. It's plotted on the same 0-100 scale as RSI but reacts more slowly because each input already spans several periods. Since it shares RSI's 0-100 scale, some traders apply RSI's overbought/oversold reference levels by analogy, though this isn't a fixed rule specific to RMI, and a slower line will cross them less often. Like any momentum oscillator, it can remain at an extreme through a strong trend, so a reading is evidence of momentum conditions, not a standalone trade signal.
The Formula
Momentum(t) = Close(t) − Close(t − n), where n is the momentum look-back, a common default is 5 periods. This replaces the single-period change (Close(t) − Close(t − 1)) that standard RSI uses as its raw input.
RMI = 100 − [100 ÷ (1 + RS)], where RS = Average of up-momentum values over the smoothing period ÷ Average of down-momentum values over the smoothing period.
In other words: compute the n-period momentum value for each bar, split those momentum values into "up" bars (positive momentum) and "down" bars (negative momentum, taken as an absolute value), average each side over the smoothing period the same way RSI averages single-period gains and losses, take the ratio as RS, and apply the same RS-to-0-100 conversion RSI uses. Because the calculation reuses RSI's averaging and smoothing conventions, exact results can vary slightly by platform depending on which smoothing method (such as Wilder's smoothing) is applied, verify the specific implementation on your charting platform before comparing values across tools.
| Step | RSI (standard) | RMI |
|---|---|---|
| Raw input per bar | Close(t) − Close(t − 1) | Close(t) − Close(t − n), commonly n = 5 |
| Up/down split | Positive vs. negative single-period changes | Positive vs. negative n-period momentum values |
| Averaging and smoothing | Same RSI averaging/smoothing method | Same RSI averaging/smoothing method, applied to momentum instead |
| Output scale | 0-100 | 0-100 |
Worked Example
Hypothetical example, for education only.
Using a 5-period momentum look-back, suppose today's close is $54.20 and the close 5 periods earlier was $51.90.
Momentum(t) = 54.20 − 51.90 = +$2.30 (an "up" momentum value for this bar)
Now suppose that over the smoothing period, the average of every "up" momentum value is $2.40 and the average of every "down" momentum value (taken as a positive number) is $1.10:
RS = 2.40 ÷ 1.10 = 2.18
RMI = 100 − [100 ÷ (1 + 2.18)] = 100 − [100 ÷ 3.18] = 100 − 31.4 = 68.6
An RMI reading of 68.6, close to but not above the 70 level some traders borrow from RSI by analogy as an overbought reference, is read together with the fact that up-momentum has been outweighing down-momentum over the smoothing window, consistent with a market where price gains, measured several periods apart, have generally been outrunning price losses. Because RMI's inputs already span multiple periods, this reading changes more gradually than a single-period RSI value would from one bar to the next.
Common RMI Signals
Overbought and oversold readings
Since RMI shares RSI's 0-100 scale, some traders apply the same 70/30 overbought/oversold zones by analogy, readings above roughly 70 described as overbought, below roughly 30 as oversold, though this isn't a fixed rule specific to RMI's own definition. Because RMI is smoother and slower, it tends to cross those levels less often than RSI on the same chart, and a strong trend can keep it pinned near an extreme for an extended stretch rather than reversing promptly. An overbought or oversold reading describes recent momentum conditions, not a guaranteed reversal.
Centerline behavior
Readings above 50 suggest up-momentum has generally been outweighing down-momentum over the smoothing period; below 50, the reverse. Because RMI reacts more slowly than RSI, a sustained move away from 50 is sometimes read as a steadier confirmation of directional momentum than a comparable single-period RSI reading.
Divergence
Bearish divergence: price makes a higher high while RMI makes a lower high. Bullish divergence: price makes a lower low while RMI makes a higher low. As with RSI, divergence flags a disagreement between price and the indicator rather than a confirmed reversal, and RMI's added smoothing means divergence patterns may take longer to appear and resolve than they would on standard RSI.
Common RMI Settings
| Momentum look-back | Smoothing | Common use |
|---|---|---|
| 5 (commonly cited default) | A period similar in spirit to standard RSI's, which commonly uses 14, though RMI's exact smoothing convention varies by platform | General-purpose momentum analysis |
The momentum look-back (how many periods back the comparison reaches) and the smoothing length (how many periods are averaged into the RS calculation) can be adjusted independently. A longer momentum look-back generally produces a smoother, later-reacting line; a shorter one moves the indicator closer to how standard RSI behaves. There's no single correct combination, verify your platform's exact default parameters and test alongside the asset and timeframe you actually trade rather than assuming the commonly cited 5-period look-back is optimal for every case.
Why RMI Produces False Signals
- Strong, persistent trends, like RSI, RMI can stay above 70 or below 30 for an extended stretch while a trend simply continues, so a threshold reading alone doesn't confirm a reversal.
- Lag from the added momentum window, comparing prices several periods apart smooths noise but also delays the line's reaction to a genuine, fresh momentum shift relative to single-period RSI.
- Sideways or choppy markets, repeated threshold crossings with little follow-through can still occur in a directionless market, even with RMI's extra smoothing.
- Comparing RMI across different look-back settings, an RMI reading computed with one momentum look-back and one computed with a different look-back on the same chart aren't directly comparable; the same principle that applies to comparing different RSI period settings applies here.
- Assuming identical behavior to RSI, because RMI shares RSI's scale and general shape, it's easy to apply RSI intuition without accounting for RMI's slower reaction time.
Common Mistakes
- Treating a 70/30 crossing as an automatic trade trigger, as with RSI, these levels describe momentum conditions, not guaranteed entry or exit points.
- Not verifying the momentum look-back and smoothing settings a platform actually uses, implementations can differ, and an unverified default can silently distort a comparison across charting tools.
- Expecting RMI to react as quickly as RSI, its added smoothing is the point, but it means signals that would already have triggered on RSI may still be forming on RMI.
- Reading divergence in isolation, divergence on RMI, like on RSI, still needs price confirmation before it's treated as more than a warning.
- Ignoring the underlying trend, a countertrend RMI signal can fail repeatedly against a strong, sustained directional move.
Limitations
RMI is derived entirely from historical closing prices, like RSI, it doesn't independently account for earnings, valuation, news, liquidity, or order-book depth. Its extra smoothing reduces some noise relative to RSI but does not eliminate false signals, and the added momentum window means it reacts to a genuine trend change later than single-period RSI would. No setting removes the underlying uncertainty of forecasting future price direction from past price behavior.
Changing What Momentum Is Measured Against
The Relative Momentum Index compares each close against a close some number of periods back rather than against the immediately preceding one. Increasing that comparison lookback smooths the resulting series and shifts the readings, so the indicator behaves differently from a conventional momentum oscillator even at the same period setting.
The practical effect is fewer, slower signals. A longer comparison window means short-lived moves do not register, which reduces false readings in choppy conditions and delays recognition of genuine turns. That trade is the whole reason to choose this variant.
The mistake is carrying over threshold levels. Because the comparison window changes the distribution of readings, the levels that identify extremes on a standard oscillator do not identify extremes here, and using them produces either constant signals or almost none depending on the direction of the shift.
The indicator also remains a momentum measure, which means readings can stay at an extreme through an entire trend. A high value indicates that recent closes have exceeded earlier ones consistently; it does not indicate that the sequence is about to end.
RMI FAQs
What is the difference between RMI and RSI?
RSI compares each close with the prior close, one period back. RMI compares each close with the close several periods back, a common default is five, before applying the same up/down averaging and RS/RSI-style smoothing formula. RMI is typically smoother and slower to react than RSI.
What is the standard RMI setting?
A commonly cited default uses a 5-period momentum look-back, then applies a smoothing period similar in spirit to standard RSI's, which commonly uses 14 periods, though RMI's exact smoothing convention varies by platform, always verify the exact defaults on your specific charting platform.
Is RMI more reliable than RSI?
Not inherently. RMI's added smoothing can reduce whipsaw in choppy conditions but also delays signals relative to RSI. Neither version removes the need for price confirmation, and which one performs better depends on the asset, timeframe, and what it's tested against.
Does RMI use the same 0-100 scale and overbought/oversold levels as RSI?
RMI is plotted on the same 0-100 scale, and since it shares that scale with RSI, some traders apply the same 70/30 overbought/oversold zones by analogy, though this isn't a fixed rule specific to RMI. Because RMI reacts more slowly, some traders adjust those thresholds or expect fewer, later signals compared with RSI on the same chart.
Who created the Relative Momentum Index?
RMI was developed by Roger Altman as a variation on Welles Wilder's RSI, substituting a multi-period momentum value for RSI's single-period price change.
Can RMI be used by itself?
It can be viewed alone, but like most momentum oscillators it works better alongside price structure, trend context, and volume rather than as a standalone trade trigger.
What does the momentum lookback add compared with a one-period change?
Comparing each price against one from several periods back, rather than against the immediately preceding one, filters out short oscillations that would otherwise register as alternating gains and losses. The result is a smoother oscillator that reaches extremes less often. The cost is a slower response, since a reversal must persist across the momentum lookback before the reading turns.
How do the two period settings interact?
One sets how far back each comparison reaches and the other sets the averaging length applied to the resulting gains and losses. Lengthening either smooths the output, but they smooth differently: the momentum lookback filters short reversals while the averaging period damps the whole series. Changing both at once makes it difficult to attribute a behaviour change to either.
Do the conventional overbought and oversold levels transfer to this oscillator?
Because the momentum lookback changes how often extremes are reached, the same threshold levels produce a different frequency of signals than they would on a one-period comparison. Levels borrowed unchanged tend to fire too rarely with a long lookback. Establishing what constitutes an extreme for the specific settings and instrument, from its own history, is more reliable than importing conventional numbers.