Chande Momentum Oscillator (CMO)

An RSI relative that uses the same up/down price changes but raw sums and a symmetric −100 to +100 scale.

What it is

The Chande Momentum Oscillator (CMO) was introduced by Tushar Chande in 1994 in The New Technical Trader. It starts from exactly the same raw material as RSI — each bar's price change split into an 'up' part and a 'down' part — but processes it differently. Where RSI smooths the gains and losses with Wilder's running average and maps the ratio onto a 0–100 scale, CMO takes the plain sums of gains and losses over the lookback window and computes the net change as a fraction of total change, scaled to a symmetric −100 to +100 range with zero as the natural midpoint. The difference between CMO and RSI is therefore pure arithmetic, not a different theory of markets: same inputs, different smoothing (none versus Wilder's) and different normalization (symmetric versus 0–100). In fact, an RSI computed with raw sums instead of Wilder smoothing relates to CMO by the exact identity CMO = 2 × RSI_raw − 100. CMO is plotted in its own pane, typically with reference lines at +50 and −50. Like all oscillators it is descriptive rather than predictive — it summarizes recent price changes and forecasts nothing. This is educational content, not financial advice, and trading involves risk of loss.

How it works

Over a lookback of N bars (default 14 or, in Chande's original work, often 9 or 20), each bar's change is classified: if the close rose, the gain is added to sumUp; if it fell, the absolute loss is added to sumDown. CMO is then 100 × (sumUp − sumDown) / (sumUp + sumDown). The numerator is the net price change over the window; the denominator is the total absolute movement, up plus down. When every bar in the window closed higher, CMO reads +100; when every bar closed lower, −100; when gains and losses balance, 0. Compare this with RSI mechanically: RSI computes avgGain / (avgGain + avgLoss) × 100 where the averages use Wilder's smoothing, an exponential-style running average that never fully forgets old data. CMO's raw sums mean every bar in the window carries equal weight and drops out abruptly after N bars — so CMO is jumpier than RSI at the same length, both because there is no smoothing and because of the drop-off effect when a large old bar exits the window. The symmetric scale also changes interpretation: CMO's zero corresponds to RSI's 50, and the conventional ±50 CMO bands correspond to raw-sum RSI readings of 75 and 25, tighter definitions of stretched momentum than RSI's usual 70/30.

How traders read it

• Scale: CMO runs from −100 to +100 with zero meaning gains and losses balanced over the window. Zero on CMO plays the role that 50 plays on RSI.
• ±50 bands: readings above +50 mean up-moves accounted for at least 75% of total movement in the window (and the mirror for −50). These describe one-sided recent movement, not a signal to act — strong trends can hold beyond ±50 for extended periods.
• Zero crossings: some traders read a cross above zero as recent momentum turning net-positive and below zero as net-negative; in ranging markets these crossings are frequent and noisy.
• Versus RSI: CMO and RSI at the same length tell a similar story, but CMO reacts faster and more sharply because it uses raw sums with no Wilder smoothing — faster is not more accurate, just less filtered.
• Divergence: price making a new extreme while CMO does not is sometimes read as movement becoming less one-sided; like all divergences it can persist for many bars without a reversal.
• Extremes: ±100 requires every bar in the window to move the same direction, so short lookbacks pin at extremes far more often than long ones.

Common settings

Length 14 is the usual platform default; Chande also worked with 9 (faster) and 20 (smoother). Reference lines at +50 and −50, sometimes with a zero midline. Source is normally the close. Some traders smooth CMO with a short moving average to tame the raw-sum jumpiness — at which point its responsiveness advantage over RSI largely disappears, which is worth understanding rather than assuming either version is 'better'.

Strengths

• Symmetric −100/+100 scale with a meaningful zero makes net direction readable at a glance.
• Transparent arithmetic: raw sums with no smoothing, so the value on any bar can be verified by hand.
• More responsive than RSI at the same length, which can be useful when you specifically want an unsmoothed reading.
• The exact identity with raw-sum RSI (CMO = 2 × RSI_raw − 100) makes it easy to reason about and cross-check implementations.
• Works on any instrument and timeframe with a single length parameter.

Pitfalls to watch

• Jumpier than RSI by construction: no smoothing means every bar's change passes straight through, producing frequent noisy swings, especially at short lengths.
• Drop-off effect: a large bar leaving the window moves CMO even when the current bar is quiet — movement in the line is not always new information.
• ±50 is a convention, not a law: markets can hold beyond ±50 through entire trends, so treating those levels as automatic reversal zones means fighting strong moves.
• Settings sensitivity: changing the length changes every reading and every band touch; a length that looked informative on one chart may not on another.
• Division-by-zero edge case: if price is completely flat across the window the denominator is zero, and implementations must guard it explicitly.
• No predictive power: CMO restates recent price changes with different normalization than RSI; neither version forecasts price, and all trading carries risk of loss.

Pine v6 example

//@version=6
indicator("CMO Example", overlay = false)

length = input.int(14, "Length", minval = 1)
src    = input.source(close, "Source")

chg     = ta.change(src)
sumUp   = math.sum(math.max(chg, 0), length)
sumDown = math.sum(math.max(-chg, 0), length)

// Guard the flat-window edge case where sumUp + sumDown = 0
cmo = (sumUp + sumDown) == 0 ? 0.0 : 100 * (sumUp - sumDown) / (sumUp + sumDown)

plot(cmo, "CMO", color = color.blue)
hline(50,  "Upper",   color = color.gray)
hline(0,   "Zero",    color = color.new(color.gray, 50))
hline(-50, "Lower",   color = color.gray)

Pro tip: When choosing between CMO and RSI, be clear about what you are actually choosing: the same up/down inputs with or without Wilder smoothing, on a symmetric versus 0–100 scale. If you smooth CMO to reduce its noise, you have largely rebuilt RSI — so pick one, understand its lag/noise trade-off, and read it alongside price structure rather than in isolation. Educational information only, not advice; no indicator removes the risk of loss.

Educational only — not financial advice, not a recommendation to trade. No indicator is predictive; trading involves substantial risk of loss.