The volatility comparison matrix

Nine quantities, all commonly called volatility, that are not the same thing. This is the fastest way to work out which one a sentence is actually talking about.

Quick answer: The volatility comparison matrix separates the nine distinct quantities that all get called volatility, showing for each what it is computed from, whether it looks backward or forward, what it actually says, its units, and what dominates it — because most disagreements between two correct volatility numbers come from comparing two different quantities.

Nine quantities, all commonly called "volatility", that are not the same thing. This table is the fastest way to work out which one a sentence — or a screener, or a broker's platform — is actually talking about.

Measured from data that exists Inferred by inverting a model Derived from other quantities Somebody's opinion about the future

Measure	Computed from	Looks	What it really says	Units	Dominated by	Kind
Historical volatility	Past prices	Backward	Realised, close-to-close	Annualised %	A window you chose	Measured
Realised volatility	Past prices (often intraday)	Backward	What actually happened	Annualised %	The estimator you chose	Measured
Implied volatility	Option prices	Forward	What the market charges	Annualised %	The model you inverted	Inferred
Expected volatility	A forecast or a model	Forward	What someone predicts	Annualised %	Whoever made the forecast	Forecast
Forward volatility	Two implied volatilities	Forward, between two future dates	The window between them	Annualised %	The term structure	Derived
India VIX	The whole NIFTY chain	Forward, fixed 30-day	Model-free expected variation	Annualised %	NSE's published method	Inferred
Variance	Any of the above, squared	Either	The additive quantity	%² (or fraction²)	The volatility it squares	Derived
IV Rank	One year of IV	Neither — it is context	Position in the 52-week range	0–100	Two extreme days	Context
IV Percentile	One year of IV	Neither — it is context	Share of days below today	0–100	Every day equally	Context

How to read this table

The Looks column is the one that causes the most damage. Historical and realised volatility are statements about a past that cannot change. Implied volatility is a statement about a future that has not happened, and it is a price, not a prediction — the number at which supply met demand. Comparing the two, as the HV vs IV page does, is comparing a photograph with a weather forecast. Both are useful; neither is the other.

The Dominated by column explains most disagreements between two correct numbers. Two people compute historical volatility on the same day and get different answers because they chose 10-day and 30-day windows. Two people look up "IV rank" and get 18 and 62 because one is reading rank and the other percentile. Neither is wrong; they are answering different questions.

The forward-looking measure and the backward-looking one, side by side

Both series describe the same underlying over the same period.

Implied volatility normally sits above trailing realised volatility — the gap is the volatility risk premium — and inverts precisely when a shock arrives.

Which one should I use?

To decide whether an option is expensive: compare implied volatility with realised volatility over a comparable window, and check IV Rank and IV Percentile together.
To size a position or set a strike: use implied volatility, because that is what the market is charging for the period you will actually be exposed to. See Expected Move.
To model a portfolio's risk: use variance, not volatility, because variance is what adds across independent positions and across time.
To describe what has already happened: use realised volatility, and say which estimator and which window.
Never compare a 10-day historical volatility with a 30-day implied volatility and conclude anything from the gap. You have compared two different periods and attributed the difference to the option market.

Frequently asked questions

How many different things are called volatility?

At least nine in common use: historical, realised, implied, expected and forward volatility, a volatility index such as India VIX, variance, and the two context metrics IV Rank and IV Percentile. They are computed from different data and answer different questions.

What is the difference between historical and realised volatility?

In practice they overlap. Historical volatility conventionally means close-to-close over a chosen window; realised volatility usually means a finer estimator, sometimes using intraday data. Both look backward at prices that already happened.

Is implied volatility a forecast?

No. It is a price — the number at which supply met demand for optionality. It is a biased forecast of subsequent realised volatility, usually exceeding it, and that bias is the volatility risk premium.

Why do two people compute different historical volatilities for the same stock?

Because they chose different windows. A 10-day and a 30-day historical volatility of the same series are different numbers and both are correct. The window is the question, not a detail.

When should I use variance instead of volatility?

When quantities must add. Variance adds across independent periods and across independent positions; volatility does not. That is why forward volatility, the square-root-of-time rule and variance swaps all exist.

Which volatility should I use to size a position?

Implied volatility, because it is what the market is charging for the period you will actually be exposed to. Use realised volatility to describe what has already happened, and variance to model a portfolio.

Last reviewed 10 July 2026. Educational content only — not investment advice.