Momentum & oscillatorsRelative Momentum Index · RMI
A momentum-smoothed cousin of RSI that compares the close to a price several bars back rather than to the prior bar.
Works best in ranging marketsEngine-computed on a fixed sample series
What the Relative Momentum Index actually measures
The Relative Momentum Index (RMI) is a momentum oscillator that Roger Altman introduced in 1993 as a smoother, more patient version of Welles Wilder's RSI. It answers the same core question as RSI — have buyers or sellers dominated the recent tape, and is the move stretched? — but it measures that pressure over a multi-bar window instead of from one bar to the next. Where RSI compares each close to the close immediately before it, the RMI compares each close to the close a fixed number of bars ago, called the momentum length. That single change filters out much of the one-bar jitter that makes RSI twitchy, giving a line that swings between 0 and 100 but hugs the extremes far less often. For a beginner, the simplest way to picture it is RSI viewed through a slightly longer lens, so its turns are calmer and its overbought and oversold readings carry more weight.
How the RMI is calculated
Begin by choosing two parameters: a momentum length M, which is how many bars back to compare, and a smoothing length N, which is the averaging window. For each bar, subtract the close M bars ago from today's close; if the result is positive it counts as an up-move of that size and the down-move is zero, and if it is negative the reverse holds. These up- and down-moves are then averaged with Wilder's smoothing over N periods, exactly as RSI averages its one-bar gains and losses. The relative momentum ratio is the average up-move divided by the average down-move, and the RMI is 100 minus 100 divided by one plus that ratio, which pins the output between 0 and 100. The elegant property is that when M equals 1 the momentum comparison collapses to a one-bar change and the RMI becomes mathematically identical to a Wilder RSI, so RSI is simply the special case of RMI with a momentum of one.
Reading the RMI line, step by step
Read the RMI on the same map as RSI: above 70 is overbought, below 30 is oversold, and the 50 line is the pivot between bullish and bearish control. Because the momentum term smooths away short-term noise, the line spends less time pinned at the extremes, so an actual print above 70 or below 30 is a more meaningful stretch than the same reading on raw RSI. The most valuable signal is divergence — price grinding to a new high while the RMI rolls over to a lower high warns that the thrust behind the move is thinning even though price has not yet turned. In a trend, watch how the RMI behaves around 50: in a healthy uptrend pullbacks often bottom in the 40 to 50 band and turn back up, so that zone acts as dynamic support for momentum. A cross of 50 from below hints the balance has tilted toward buyers, and a cross from above tilts it toward sellers.
Best timeframes and settings
The RMI is a swing- and position-trading tool at heart, and it shines on daily and 4-hour charts where its extra smoothing separates real exhaustion from ordinary noise. Altman's common configuration is a smoothing length near 20 with a momentum of 4 or 5, though many platforms default to a 14-length with a momentum of 3 to 5. Raising the momentum length makes the line smoother and slower, which is good for filtering a whippy market but at the cost of later turns, while lowering it toward 1 pushes the RMI back toward plain, jumpy RSI. Shorter smoothing lengths make it more reactive for intraday work but reintroduce false extremes. The practical trade-off is always the same: more smoothing buys you cleaner divergences and fewer fake signals, but you pay for it in lag at sharp reversals, so match the settings to how much noise your instrument and timeframe throw off.
When and where the RMI works
The RMI is at its best in markets that oscillate — rangebound stocks, mean-reverting FX pairs, and consolidations — where fading its overbought and oversold turns has an edge. It also works as a momentum gauge inside a trend, using the 40 to 60 band as a pullback zone to buy dips or sell rallies with the prevailing direction. Where it struggles, like every bounded oscillator, is in a powerful one-way trend: the added smoothing helps, but the RMI can still park in overbought for weeks while a strong uptrend runs, so a lone overbought reading is never a short. Avoid trading its extremes blindly during news-driven breakouts or the opening of a fresh trend. It suits liquid instruments across equities, futures, and currencies, and is less reliable on thin, gappy tickers where the momentum comparison is distorted.
Strategies built on the RMI
A range-fade strategy waits for the RMI to push above 70 and then hook back down through it, entering short as it re-crosses, with a stop above the swing high and a target near the range midline or the opposite band; the mirror applies for longs off oversold. A trend-pullback strategy first confirms an uptrend, with price above a rising 50- or 200-period average, then buys when the RMI dips into the 40 to 50 zone and turns up, stopping below the recent swing low and trailing as the trend extends. A divergence strategy is the most selective: mark a price high that is not confirmed by a new RMI high, wait for a break of the minor swing low or a 50-line cross to confirm the momentum failure, then enter against the tiring move with a stop beyond the extreme. In every case the RMI times the entry, but the direction should be set by structure or a trend filter, not by the oscillator alone.
Combining the RMI with other tools
The RMI pairs naturally with a trend filter such as the 200-day SMA or a MACD zero line, which tells you whether to fade its extremes in a range or only trade its dips with the trend. Volume or a volume-based tool like OBV can confirm whether a divergence has real distribution behind it. Support and resistance or pivot levels give the RMI's overbought and oversold turns a price reason to reverse, sharply improving the hit rate over trading the oscillator in a vacuum. Because it belongs to the RSI family, it also complements rather than duplicates a differently constructed oscillator like the stochastic or the Stochastic Momentum Index, and agreement between two independent momentum reads is a stronger tell than either alone.
Where the RMI fails
The classic failure is treating an overbought RMI as a sell in a strong uptrend; the smoothing delays the extreme but does not stop the line from riding high while price keeps climbing. A large momentum length adds real lag, so at a sharp V-reversal the RMI can turn several bars after price, handing back part of the move. In dead, choppy markets the line can still oscillate across 50 repeatedly, producing whipsaw crosses with no follow-through. Divergence, its best feature, is also a trap: momentum can diverge for many bars before price finally responds, so acting on the first divergence without a confirming trigger often means being early and stopped out. The fix in each case is confirmation — a trend filter, a structural level, or a price trigger — never the RMI's reading on its own.
A worked example
Imagine a stock in a broad range using a 14-length RMI with a 5-bar momentum. Over the smoothing window the average up-move, each measured against the close five bars earlier, works out to 2.0 points and the average down-move to 0.8 points. The relative momentum ratio is 2.0 divided by 0.8, or 2.5, so the RMI reads 100 minus 100 divided by 3.5, which is about 71 — just into overbought. Price then tags the top of its range and the next bar closes lower, dragging the RMI back below 70. A range-fader shorts that re-cross at 118.50, places a stop at 121 above the swing high, and targets the range midpoint near 112. Notice the RMI did not tell you to short at 71 alone; it was the combination of a stretched reading, a known resistance level, and the hook back below 70 that formed the actual signal.