Stochastic Momentum Index (SMI)

A double-smoothed stochastic that measures where the close sits relative to the midpoint of the recent range, producing a centered oscillator from -100 to +100.

What it is

The Stochastic Momentum Index (SMI) was introduced by William Blau in a 1993 Technical Analysis of Stocks & Commodities article as a refinement of the classic stochastic oscillator. The difference between the two is exact and worth stating precisely, because it answers the most common question about SMI. The regular stochastic asks: where is the close relative to the low of the recent high-low range? Its answer runs from 0 (at the range low) to 100 (at the range high), so its natural center is an awkward 50. The SMI instead asks: where is the close relative to the midpoint of that same range? Its answer is naturally centered at zero — positive when the close is in the upper half of the range, negative in the lower half — and Blau scales it to roughly -100 to +100. He then applies double exponential smoothing to both the numerator and denominator, which makes the line considerably smoother than a raw %K. The result is a centered, smoothed range-position oscillator. Like the stochastic it descends from, SMI describes where price has recently closed within its recent range; it predicts nothing. This entry is educational only, not financial advice, and trading involves risk of loss.

How it works

The calculation runs in four steps. First, over a lookback of k bars (commonly 10), find the highest high and lowest low, and compute the range midpoint: (highest high + lowest low) / 2. Second, form two raw series: the distance of the close from that midpoint (close minus midpoint), and the full range width (highest high minus lowest low). Third — the step that defines SMI — smooth each of those series twice with EMAs (commonly 3-period, applied twice). Smoothing numerator and denominator separately before dividing is deliberate: it damps the bar-to-bar jitter that makes raw %K so jagged, while keeping the ratio meaningful. Fourth, divide the smoothed distance by half of the smoothed range and multiply by 100. The half-range divisor is what produces the ±100 scale: a close pinned at the very top of the range sits a half-range above the midpoint, giving +100; a close at the very bottom gives -100; a close exactly at the midpoint gives 0. A signal line — an EMA of the SMI itself — is usually added. Note the divisor can approach zero when the range is extremely narrow (flat markets), so implementations must guard the division. Every component is built from completed past bars, so the SMI lags by construction, and the double smoothing that makes it clean also makes it slower than the stochastic it refines.

How traders read it

• Zero line: SMI above zero means smoothed closes have been in the upper half of the recent range; below zero, the lower half. The zero cross is the natural center reference, unlike the stochastic's 50.
• Extremes: readings beyond roughly +40 or -40 are conventionally described as stretched — the close has been persistently near the top or bottom of its range. In a strong trend, SMI can hold extreme readings for many bars, and that persistence describes trend strength, not an imminent reversal.
• Signal-line crossings: SMI crossing its EMA signal line is read the same way as any oscillator/signal pair — a change in the smoothed reading's direction, observed after the fact.
• Compared with stochastic: on identical data, SMI turns more smoothly and crosses its center less often than raw %K crosses 50 — fewer flips, but each one confirmed later. That is the trade, not an improvement for free.
• Divergence: price making a new high while SMI makes a lower high is read by some as range-position momentum fading; as with all divergence, it can persist indefinitely and is not a timing tool.
• Very narrow ranges make the reading unstable — when the denominator is tiny, small price moves swing SMI violently, so readings in dead-flat conditions deserve skepticism.

Common settings

Common defaults are a 10-bar range lookback (%K length), double EMA smoothing of 3 and 3, and a 3-period EMA signal line; reference lines at +40 and -40 with a zero midline. Blau's own writings use several combinations (e.g., longer lookbacks with heavier smoothing such as 25/13). Longer smoothing gives a cleaner but slower line; shorter values approach the choppiness of the raw stochastic. As always, the defaults are conventions to test, not discovered truths.

Strengths

• Naturally centered at zero, which makes above/below-midline reading cleaner than the stochastic's 50-line convention.
• Double smoothing of numerator and denominator separately produces a markedly smoother line than raw %K without simply lagging a finished %K.
• Bounded ±100 scale keeps readings comparable across instruments and timeframes.
• The math is fully specified and reproducible — midpoint distance over half-range, double-EMA smoothed — so platform implementations can be verified exactly.
• A good side-by-side teaching case: plotting SMI against the classic stochastic on the same data shows precisely what smoothing and re-centering buy and what they cost.

Pitfalls to watch

• Extra lag is the price of smoothness: two rounds of EMA smoothing mean SMI confirms turns later than the raw stochastic — the cleaner line is not earlier information, it is later information presented more calmly.
• Extreme readings are not reversal signals: in trends, SMI can sit beyond +40 or below -40 for long stretches; treating those levels as automatic fade points means fighting the trend.
• Division-by-near-zero instability: when the recent range is very narrow, the denominator shrinks and SMI can whip between extremes on trivial price moves; unguarded implementations can misbehave.
• Settings sensitivity: lookback and both smoothing lengths all change every reading and crossing; a combination that looks clean on one chart is curve-fit until tested elsewhere.
• Whipsaw remains in ranges: smoothing reduces but does not eliminate back-and-forth signal-line crossings in sideways markets.
• No predictive power: SMI describes where price has closed within its recent range, nothing more. It does not forecast price, and no indicator removes the risk of loss.

Pine v6 example

//@version=6
indicator("Stochastic Momentum Index (SMI)", overlay = false)

lengthK = input.int(10, "Range Lookback (%K)", minval = 1)
smooth1 = input.int(3,  "First EMA Smoothing", minval = 1)
smooth2 = input.int(3,  "Second EMA Smoothing", minval = 1)
sigLen  = input.int(3,  "Signal EMA Length", minval = 1)

hh  = ta.highest(high, lengthK)
ll  = ta.lowest(low, lengthK)
mid = (hh + ll) / 2

distSm  = ta.ema(ta.ema(close - mid, smooth1), smooth2)
rangeSm = ta.ema(ta.ema(hh - ll, smooth1), smooth2)

smi = rangeSm != 0 ? 100 * distSm / (rangeSm / 2) : 0.0
sig = ta.ema(smi, sigLen)

plot(smi, "SMI", color = color.blue)
plot(sig, "Signal", color = color.orange)
hline(40, "Upper", color = color.gray)
hline(0, "Zero", color = color.new(color.gray, 50))
hline(-40, "Lower", color = color.gray)

Pro tip: Plot SMI and the classic stochastic on the same chart with matching lookbacks and watch a few dozen turns: you will see SMI's crossings arrive later but flip far less often. Whether that trade suits you depends on what you are reading it for — there is no setting where you get the smoothness without the delay. If your platform's SMI values differ from another's, check the smoothing type (EMA vs other) and order of operations before assuming a bug. Educational context only, not advice; SMI is descriptive, and all trading carries risk of loss.

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