Standard Deviation

If you've ever described a market as "calm" or "wild" and wanted a number behind that feeling, standard deviation is it.

In 30 Seconds

• Measures spread — how far readings typically sit from their average.
• Small = calm, large = volatile — it's the core measure of variability.
• The unit inside a z-score — "two standard deviations above average" is a z-score of +2.
• Depends on its window — spread is always measured over some stretch of history.
• HIE discloses exactly how it's computed — sample standard deviation, strictly-trailing, window shown.

Definition

Standard deviation answers: on average, how far do readings fall from their average? If a market's daily moves cluster tightly around zero, its standard deviation is small. If they swing widely in both directions, its standard deviation is large. It's expressed in the same units as the data itself — points, dollars, contracts — so a standard deviation of "20 points" means readings typically land about 20 points from their average. (This point is easy to miss: measure price in points and its standard deviation is in points; measure volume in contracts and its standard deviation is in contracts.)

"Standard deviation is the typical distance of readings from their average."

It's built from the average in two steps in concept: measure how far each reading is from the average, then combine those distances into a single typical value. The larger that typical distance, the more variable — the more "spread out" — the data is.

Standard deviation is the foundation behind several other important statistical measures. A z-score is simply "how many standard deviations from average" a value is. Volatility, in quantitative finance, is commonly measured as the standard deviation of returns. So understanding standard deviation is the key that unlocks a whole family of measures built on top of it.

Why It Matters

Variability is often more important than the average. Two markets can have the same average daily move — say, near zero — while one drifts quietly and the other lurches violently. The averages are identical; the experience of trading them is completely different, and standard deviation is what tells them apart. A number's average tells you the center; its standard deviation tells you how much to expect it to stray from that center.

This is why standard deviation underlies so much of market analysis. "Is this move large?" only has meaning relative to how spread out moves normally are — a 30-point move is enormous when the standard deviation is 10 and trivial when it's 60. That relative judgment is exactly what a z-score performs, by dividing the move by the standard deviation. Without a measure of spread, "large" and "small" are just guesses.

To ground it with no math: picture two weeks in a market. In the first, daily moves are −5, +4, −3, +6, −2 — small, tightly clustered, a low standard deviation. In the second, they're −45, +60, −30, +55, −50 — huge swings in both directions, a high standard deviation. Both weeks might average out to roughly the same small number, but the second week's standard deviation is many times larger. That single number captures "how wild was it" in a way the average completely misses.

How HIE Computes It

Two platforms can report different standard deviations for the same market, because they may measure over different windows or use a different formula (sample vs. population). That's why HIE discloses its methodology rather than presenting a bare number.

HIE computes standard deviation with a single disclosed, deterministic method — and every answer prints the exact window it used. Three things define it:

• Sample standard deviation (ddof = 1). HIE divides by n−1 rather than n — the standard choice when computing spread from a *sample* of history rather than a complete population. It's the honest default when you're estimating variability from observed data.
• Strictly-trailing window. The spread is measured against a value's own trailing history only — the calculation at any bar uses nothing from after that bar. That is what "no look-ahead" means.
• A disclosed default window per feature, which you can override. Each feature carries its own default from HIE's registry (100 bars for price and RSI, 20 for volume and ATR); when you don't state a window, HIE asks (default, 250, 500, or your own). The window used is always printed.

Standard deviation is a live, standalone function — you can ask for it directly, and the same computation also powers every z-score. In HIE's disclosure ("what ran"), a standalone standard-deviation comparison appears in this form:

VOLUME > STD(VOLUME, window=20, method=sample SD (ddof=1), rolling, strictly-trailing, v1)

That same `method=sample SD (ddof=1)` also appears inside every z-score's disclosure — deliberately, because there is one definition of standard deviation in the whole engine. Whether you ask for standard deviation directly or ask for a z-score built on it, the underlying computation is identical. The shorter reference form is:

STD(<FEATURE>, window=<N>, method=sample SD (ddof=1), rolling, strictly-trailing, v1)

In the engine's function grouping, standard deviation is the *Dispersion* function (alongside the Normalization functions z-score and percentile, and the Aggregation functions mean, median, max, and min).

Interpreting Standard Deviation

Standard deviation is read relative to the data's own scale — there's no universal "high" or "low," because it's in the units of whatever you're measuring:

• Small standard deviation — readings cluster tightly around the average; a calm, orderly, low-variability period.
• Large standard deviation — readings scatter widely; a volatile, high-variability period.
• Rising standard deviation — variability is increasing; the market is becoming more turbulent.
• Falling standard deviation — variability is contracting; the market is calming (often called compression).

A useful rule of thumb *if* data were normally distributed: roughly two-thirds of readings fall within one standard deviation of the average, and about 95% within two. But markets have fatter tails than a normal distribution, so extreme readings beyond two or three standard deviations happen more often than that rule suggests — a caution that carries straight over to z-scores built on standard deviation.

Common Mistakes

"A high standard deviation is bad."

Neither good nor bad on its own — it's a description of variability. High standard deviation means a volatile period; whether that's good or bad depends entirely on what you're doing. It's a measurement, not a verdict.

"Standard deviation predicts the next move."

No. It summarises how spread out past readings were. It tells you how much variability there has been, not what happens next.

"There's one true standard deviation for this market."

No. Standard deviation depends on the window it's measured over and whether the sample or population formula is used. Measured over 20 bars versus 200 bars, you'll get different numbers — both correct for their window. This is why HIE discloses the window and method.

"Sample and population standard deviation are the same thing."

Close, but not identical. The sample version (dividing by n−1) is slightly larger and is the correct choice when estimating spread from a sample of history — which is what HIE uses. On large windows the difference is tiny; on small ones it matters.

Limitations

Standard deviation is only as meaningful as its window and the data's shape. Over a short trailing window, it's estimated from few readings and can jump around on little evidence — the same small-sample caution that applies to every statistic.

Its deeper limitation is that standard deviation summarises spread with a single symmetric number, which fits data that's roughly balanced around its average but describes lopsided (skewed) data poorly. Market returns are often skewed and fat-tailed, so standard deviation can understate how often extreme moves occur, and it treats an unusually large up-move and down-move as equally "far" even when the market doesn't. It's an essential, foundational measure — but it's a summary, and like every summary it hides some of what it compresses.

Try These HIE Questions

Standard deviation is a live, askable function — you can ask whether a feature is above or below its own recent standard deviation, or ask for a z-score built on it. This query runs in HIE today and returns a certified answer:

• "ES volume above its standard deviation" (default 20-bar window for volume) — resolves to `VOLUME > STD(VOLUME, window=20, method=sample SD (ddof=1), rolling, strictly-trailing, v1)`.

*(A real answer generated on 17 July 2026, research through that date; because the historical data is periodically refreshed, the exact counts can change over time.)*

The window here — 20 bars — is volume's registry default; you didn't have to choose it, and HIE disclosed it anyway. Standard deviation applies to price, volume, RSI, and ATR alike, and the same sample-SD computation powers every z-score.

How This Fits Into MPM

Standard deviation is the foundational variability measure beneath much of HIE's work — most directly, it's the unit inside every z-score, and the basis for how HIE expresses "unusual" in a way that adapts to changing volatility.

You'll encounter it in:

• HIE (Historical Intelligence Engine), where standard deviation is computed as sample SD over a disclosed, strictly-trailing window and appears in the z-score disclosure string, so the spread behind any "unusualness" judgement is never hidden.
• Research Papers, where variability and "unusualness" are expressed in disclosed standard-deviation (or z-score) units rather than raw points, so a claim holds up across calm and volatile eras.
• Intelligence Circle — a member-only research environment containing advanced market research, historical investigations, and trading strategies developed using the MPM research framework.

MPM's broader stance is that a measure of variability is only honest if its window and formula are disclosed — which is why HIE states both rather than presenting an unexplained number.