Core Volatility Intermediate Dispersion within a single session Backward-looking

Intraday Volatility

Everything a closing price throws away, measured minute by minute.

Quick answer: Intraday volatility is the dispersion of an asset's returns measured within a single trading session, revealing a U-shaped pattern — violent at the open, quiet at midday, rising into the close — that a volatility computed from closing prices never sees.

In simple words

Intraday volatility is how much a price moves around during the day, measured in slices of the session rather than from one day's close to the next. It is not the same in every slice. On the NSE, the first few minutes after the 9:15 open are wild — the price is catching up on everything that happened overnight, when the market was shut but the world was not. Things settle through the morning, go quiet around lunch, and pick up again into the 15:30 close as traders square off positions they do not want to hold overnight. Plot the volatility of each slice against the time of day and you get a U-shape: high at both ends, low in the middle.

The important consequence is that a volatility computed only from closing prices is blind to all of this. Close-to-close volatility sees one number per day and cannot tell you that the day's movement was concentrated in the first fifteen minutes and the last thirty. Two days can have identical closing-to-closing moves — one a smooth drift, the other a violent open that fully reversed by lunch and a frantic close — and close-to-close volatility calls them the same. For anyone who has to trade during the session, that hidden shape is often the whole story.

Not to be confused with: Overnight gap risk, which is a separate distribution entirely. Intraday volatility measures movement while the market is open; the gap between one day's close and the next day's open is movement that happened while it was shut, driven by overnight news, and it follows different statistics. Close-to-close volatility silently blends the two into one number, which is why it can mislead about both.

The shape of a trading day, minute by minute

The U-shaped session: loud at the open, quiet at lunch, loud into the close

Average volatility of NIFTY by intraday bucket, from the 9:15 open to the 15:30 close.

The first bucket after the open is roughly three times as volatile as a mid-session bucket, volatility falls through the morning to a midday trough, and it rises again into the close. The picture proves that risk is not spread evenly across the session, so a stop, a fill or a hedge placed at 9:20 is exposed to a completely different volatility from the same order placed at 12:30 — and a single close-to-close number averages all of it into invisibility.

Professional explanation

The U-shape, and why the open is the loudest bucket

The defining fact of intraday volatility is its shape. The first bucket after the 9:15 NSE open is roughly three times as volatile as a bucket in the middle of the session, and the reason is mechanical: while the market was closed, information did not stop arriving. Overnight moves in US and Asian markets, currency shifts, company announcements, global news — all of it accumulates with no price to absorb it, and at 9:15 the price has to reprice for the entire gap in a compressed burst of trading. Volatility then decays through the morning as that backlog clears, reaches a trough around midday when the least new information is arriving, and turns back up into the afternoon. The U-shape is not a quirk of one market; it appears in essentially every exchange-traded market with a fixed open and close.

The close is loud for a different reason than the open

The right-hand arm of the U rises for reasons that have nothing to do with overnight news. Into the 15:30 close, volatility increases because participants are squaring positions they do not want to carry overnight, because the day's directional views are being expressed before the book flattens, and because closing prices matter — they set settlement values, mark-to-market levels and benchmark fixings, so there is genuine demand to trade at or near them. The open reprices accumulated information; the close concentrates position-management and benchmark trading. Both produce high volatility, but they are different animals, and a model that treats the two ends of the U as the same phenomenon misunderstands both.

Overnight gap risk is a separate distribution

The most consequential thing intraday volatility teaches is that the trading day and the overnight gap are two different risks with two different distributions. Intraday movement is continuous — you can watch it, react to it, hedge it, place a stop inside it. The overnight gap is discontinuous: the market closes at one price and opens at another, with no opportunity to trade in between, so a stop-loss offers no protection against it. Gap risk has fatter tails and is driven by scheduled overnight events and global moves, while intraday volatility is driven by the flow of the session itself. Close-to-close volatility adds these two distinct sources together and reports a single standard deviation, which describes neither well — it understates the tail risk of the gap and averages away the shape of the session.

Why a close-to-close number cannot see any of this

Close-to-close volatility, the standard historical measure, samples the price once a day. By construction it cannot observe anything that happens between those samples. It cannot tell you that Monday's move all happened in the first fifteen minutes, that Tuesday's close-to-close calm concealed a violent open that fully reversed, or that most of a week's realised range was overnight rather than intraday. Range-based estimators — Parkinson, Garman–Klass — recover some of this by using the day's high and low, or its open, high, low and close, and are several times more statistically efficient than close-to-close for exactly that reason. But even they compress the session into a few numbers. To see the U-shape itself you have to sample within the day, bucket by bucket, and accept that the resulting picture is far richer and far noisier than a single daily figure.

Expiry day on NIFTY weeklies is its own animal

The general U-shape describes an ordinary session. NIFTY weekly expiry days break it. As a weekly option approaches its final hours, the gamma of at-the-money options explodes: near expiry, a small move in NIFTY produces a large change in an option's delta, so hedgers must trade increasingly aggressively to stay neutral, and their hedging itself amplifies the moves. Intraday volatility on expiry afternoons can spike far above a normal close, pin the index to a strike where the most open interest sits, or whip it violently as that pin breaks. The at-the-money premium is collapsing toward zero at the same time, so the volatility implied by those tiny premiums becomes wildly unstable. Anyone measuring or trading intraday volatility on a NIFTY weekly expiry is looking at a distribution that does not resemble any other session of the week.

Formula

Annualising an intraday volatility

σ_annual = σ_bucket × √(B × 252)

To put a per-bucket volatility on the standard annual scale, scale it up by the square root of the number of buckets in a session times the number of trading days in a year. On the NSE the 9:15-to-15:30 session is 375 minutes, so 5-minute buckets give B = 75. The rule assumes buckets are independent, which the U-shape and intraday autocorrelation violate — so treat the annualised figure as indicative, not exact.

σ_annualThe annualised volatility implied by the intraday sample — the figure comparable to India VIX and to daily-sampled volatility.
σ_bucketThe volatility (standard deviation of returns) measured over a single intraday bucket, such as one 5-minute interval.
BThe number of buckets in one trading session. For 5-minute buckets in a 375-minute NSE session, B = 75.
252The number of trading days in a year, the same convention used to annualise any volatility.
√(B × 252)The scaling factor from one bucket to one year — the square root of the total number of buckets in a year. For B = 75 it is √18900 ≈ 137.5.

Parkinson's range-based intraday estimator

σ²_Parkinson = (1 / (4·ln2)) × (ln(H / L))²

A single-day variance estimate from the session's high and low alone, far more efficient than a close-to-close estimate because the high-low range uses information the two closing prices throw away. H is the session high, L the session low, and ln2 the natural log of 2 (≈0.693); the 1/(4·ln2) constant makes it an unbiased estimate of variance under a continuous random walk with no drift.

How to measure and annualise intraday volatility

Choose a bucket size — 5 minutes is common on the NSE — and divide the 9:15-to-15:30 session into equal intervals, giving 75 five-minute buckets.
Compute the return in each bucket (log of the ratio of bucket-close to bucket-open) across many sessions.
Take the standard deviation of returns within each bucket position across days — all the first buckets together, all the second buckets together — to reveal how volatility varies by time of day.
Plot bucket volatility against time of day. You should see the U-shape: the opening bucket several times the midday trough, rising again into the close.
Handle the open separately. The first bucket contains repriced overnight information and is not comparable to a mid-session bucket, so do not let it distort an average you intend to treat as typical.
To annualise a representative bucket volatility, multiply by √(B × 252) — for 5-minute buckets, √(75 × 252) = √18900 ≈ 137.5 — while remembering the independence assumption is violated intraday.
Flag expiry days. NIFTY weekly expiry afternoons follow a different distribution driven by option gamma, and mixing them into an ordinary-session average corrupts both.

Practical example

NIFTY worked example

NIFTY at 24,000. Suppose you sample the session in 5-minute buckets and find that a typical mid-session bucket has a return standard deviation of about 0.10%. There are 75 five-minute buckets in the 375-minute NSE session, so to annualise you multiply by √(75 × 252) = √18900 ≈ 137.5: the annualised volatility is 0.10% × 137.5 ≈ 13.7%, reassuringly close to India VIX near 13. Now look at the opening bucket, which runs roughly three times as volatile — about 0.30%. If you naively annualised that first bucket the same way you would get 0.30% × 137.5 ≈ 41%, a figure that is real for those five minutes and absurd as a description of the day. Interpret it: the open genuinely carries triple the volatility of midday, but that intensity does not persist, so annualising a single opening bucket overstates the session enormously. The number you can compare to VIX comes from a representative bucket, not the loudest one.

BANKNIFTY worked example

BANKNIFTY at 52,000 makes the expiry-day warning concrete. On an ordinary session BANKNIFTY's 5-minute buckets might show a mid-session standard deviation of about 0.13%, annualising to 0.13% × 137.5 ≈ 17.9% — higher than NIFTY, as expected for a bank-heavy index. But on a weekly expiry afternoon the picture distorts beyond recognition: as at-the-money options approach zero time value, their gamma spikes, hedgers trade ever more aggressively to stay neutral, and their own hedging whips the index around the strike holding the most open interest. A mid-afternoon expiry bucket can run several times its normal size, and the volatility implied by the collapsing premiums swings wildly on tiny price changes. Averaging expiry-afternoon buckets into an ordinary-session estimate would inflate the whole figure; the lesson is that intraday volatility is not one distribution but several, and expiry is a regime of its own that must be measured separately.

Lot sizes used above (NIFTY 75, BANKNIFTY 30) are those in force at the time of writing; NSE revises them periodically. Figures exclude brokerage, STT, exchange charges, stamp duty and GST. Examples are teaching scenarios built on round numbers — they are not historical quotes, not backtests and not trade calls.

Risk note. Intraday volatility invites over-leverage because the midday trough makes the market look calm precisely when it is cheapest to build a large position — and that position is then carried into the loud close and, worse, across the overnight gap that no intraday stop can protect against. A stop-loss set during the quiet midday session offers no defence against a gap open the next morning, and the calm that made the trade feel safe is a feature of one bucket, not of the risk being taken.

Advantages & limitations

What it is good for

It reveals the shape of risk within the day, showing that a stop, fill or hedge at 9:20 faces a completely different volatility from the same order at 12:30 — information a single daily number erases.
It separates intraday movement from overnight gap risk, two distinct distributions that close-to-close volatility blends into one misleading standard deviation.
It supports far more efficient estimation. Range-based estimators like Parkinson and Garman–Klass use the session's high, low, open and close to estimate daily volatility several times more precisely than a close-to-close figure from the same number of days.
It is essential for execution and timing. Anyone routing large orders uses the U-shape to schedule trading away from the volatile open and into deeper, calmer liquidity, reducing market impact.
It exposes regime differences a daily measure hides, such as the distinct volatility signature of a NIFTY weekly expiry afternoon versus an ordinary session.

Where it breaks down

It is noisy. Sampling within the day produces far more data but also far more microstructure noise — bid-ask bounce, discrete ticks, uneven liquidity — so a raw intraday estimate can overstate true volatility unless the noise is handled.
Its buckets are not independent, which breaks the annualisation. The U-shape and intraday autocorrelation mean √(B × 252) scaling is only indicative; a bucket-derived annual figure should never be treated as an exact equivalent of a daily-sampled one.
It says nothing about the overnight gap. By definition intraday volatility measures only the open session, so it captures none of the discontinuous, fat-tailed risk between one close and the next open, which is often the larger danger.
It is regime-dependent within the day itself. The opening bucket, the midday trough, the closing ramp and a weekly expiry afternoon are effectively different distributions, so a single intraday volatility figure conceals as much as it reveals unless the time of day is specified.
It is sensitive to bucket size. Choosing 1-minute versus 5-minute versus 15-minute buckets changes the estimate, because finer buckets pick up more microstructure noise and coarser ones smooth away the very shape you are trying to see.

Common mistakes

Annualising the opening bucket as if it were typical. The first bucket after 9:15 is roughly triple the midday volatility, so scaling it up implies an absurd 40%-plus annual figure that describes five minutes, not the session.
Trusting a midday calm and sizing a position against it. The quiet trough is the bottom of the U, not the risk of the day; that position is carried into a loud close and across an overnight gap no intraday stop can protect against.
Treating a close-to-close volatility as a description of the session. A single daily number cannot tell you whether the move happened at the open, at the close, or overnight, and two identical daily moves can have wholly different intraday shapes.
Blending overnight gaps into an intraday estimate, or vice versa. The gap is a separate, fatter-tailed distribution; mixing it into a within-session figure corrupts both, understating the gap's tail and inflating the session's body.
Averaging expiry-day buckets into an ordinary-session estimate. NIFTY weekly expiry afternoons are gamma-driven and follow their own distribution, so including them inflates a figure meant to describe a normal session.
Ignoring microstructure noise at fine bucket sizes. Sampling every minute picks up bid-ask bounce that inflates the volatility estimate; a raw 1-minute figure can overstate true volatility unless the noise is explicitly accounted for.

Professional usage

Execution desks live and die by the U-shape. An algorithm routing a large NIFTY order schedules its participation around the intraday volatility curve — trading less into the violent open, more into the deeper, calmer midday and late-session liquidity — because trading when volatility and spreads are widest maximises market impact and slippage. Volume-weighted and time-weighted execution strategies are built directly on the empirical shape of the session, and a desk that mis-estimates the curve pays for it in worse fills on every order. The open and close are also where the desk expects, and budgets for, the most impact.

Volatility researchers and market makers use intraday data to build realised-volatility estimators far more accurate than close-to-close, aggregating high-frequency squared returns and range-based measures into a daily figure that converges on true volatility much faster. Option market makers on NIFTY weeklies pay special attention to intraday gamma dynamics on expiry afternoons, because their hedging demand is itself a driver of the volatility they are trying to hedge, and misjudging the expiry-day regime — where a pin to a high-open-interest strike can hold or break violently — is a direct route to a hedging loss.

Key takeaways

Intraday volatility is the dispersion of returns within a single session, and it is not constant — it follows a U-shape, high at the 9:15 open, low at midday, rising into the 15:30 close.
The opening bucket is roughly three times as volatile as a midday bucket because the price is repricing all the information that arrived while the market was shut.
Overnight gap risk is a separate, fatter-tailed distribution that no intraday stop can protect against, and close-to-close volatility silently blends it with intraday movement.
A volatility computed from closing prices sees none of the session's shape; range-based estimators recover some of it, but only within-day sampling reveals the U itself.
NIFTY weekly expiry afternoons are a distinct gamma-driven regime whose intraday volatility does not resemble any ordinary session and must be measured separately.

Intraday volatility is the risk a closing price forgets to mention. The market does not move in equal instalments through the day — it reprices the world in the first fifteen minutes, dozes at lunch, and squares up into the close — and it saves its most treacherous behaviour for the gap you cannot trade and the expiry afternoon that follows no ordinary rules. A single daily volatility number is a convenient summary that averages all of this away. If you actually have to place an order, set a stop or hedge a book inside the session, the shape is not a detail; it is the terrain, and the trader who cannot see it is navigating by a number that was never meant to describe where the danger is.

Frequently asked questions

What is intraday volatility?

Intraday volatility is the dispersion of an asset's returns measured within a single trading session, in slices of the day rather than from close to close. It reveals that volatility is not constant through the day — it is high at the open, low at midday, and rises into the close, forming a U-shape.

Why is the market most volatile at the open?

Because while the market was closed, information did not stop arriving — overnight global moves, currency shifts, company news. At the 9:15 NSE open the price has to reprice all of it at once in a burst of trading, which makes the first bucket roughly three times as volatile as a mid-session bucket.

Why does intraday volatility form a U-shape?

Volatility is high at the open because the price is absorbing accumulated overnight information, falls through the morning to a midday trough when the least new information arrives, and rises into the close as traders square positions and benchmark trades cluster. Plotted against time of day, that pattern is a U.

Why is volatility high into the close?

Because participants square positions they do not want to hold overnight, express the day's directional views before flattening their books, and trade around the closing price, which sets settlement and benchmark levels. The close is loud for position-management reasons, unlike the open, which is loud because of overnight news.

What is overnight gap risk?

Overnight gap risk is the movement between one day's close and the next day's open, when the market is shut and cannot be traded. It is a separate, fatter-tailed distribution from intraday movement, and no intraday stop-loss can protect against it, because there is no opportunity to trade through the gap.

Why can't close-to-close volatility see intraday patterns?

Because it samples the price only once a day, so it cannot observe anything between those samples. It cannot tell whether the day's move happened at the open, at the close, or overnight, and two days with identical close-to-close moves can have completely different intraday shapes.

How do I annualize an intraday volatility?

Multiply the per-bucket volatility by the square root of the number of buckets in a year. For 5-minute buckets in the 375-minute NSE session there are 75 buckets a day, so the factor is √(75 × 252) = √18900 ≈ 137.5. The independence assumption is violated intraday, so treat the result as indicative.

How many 5-minute buckets are in an NSE trading day?

The NSE equity session runs from 9:15 to 15:30, which is 375 minutes, giving 75 five-minute buckets. That count is what you use to annualise a 5-minute bucket volatility, via the square root of 75 times 252 trading days.

What are Parkinson and Garman–Klass estimators?

They are range-based volatility estimators that use the session's high and low — and, for Garman–Klass, the open and close as well — instead of just the closing price. Because the high-low range carries information two closing prices discard, they estimate daily volatility several times more efficiently than a close-to-close figure.

Why is NIFTY expiry-day intraday volatility different?

Because as weekly at-the-money options approach zero time value, their gamma spikes, so hedgers must trade ever more aggressively to stay neutral, and that hedging amplifies the moves. The result is a distinct regime — pinning to a high-open-interest strike or whipping violently — that does not resemble an ordinary session.

Can a stop-loss protect me against a gap?

No. A stop-loss can only execute while the market is open, so it offers no protection against an overnight gap, where the price jumps from one close to the next open with no trading in between. Gap risk is precisely the risk a stop cannot manage, which is why it must be sized separately.

Is intraday volatility higher or lower than daily volatility?

Neither directly — they are measured differently. A per-bucket intraday figure is small in absolute terms but annualises to roughly the same neighbourhood as daily-sampled volatility if you use a representative bucket. The point of intraday volatility is not a different level but the shape and timing a daily number hides.

What bucket size should I use for intraday volatility?

It depends on the trade-off. Finer buckets like 1-minute reveal more detail but pick up more microstructure noise that inflates the estimate; coarser buckets like 15-minute are cleaner but smooth away the U-shape. Five minutes is a common compromise on the NSE for seeing the session's shape without excessive noise.

Does intraday volatility include the overnight move?

No. By definition it measures only the open session, from 9:15 to 15:30, so it excludes the overnight gap entirely. That is exactly why it must be paired with a separate estimate of gap risk — each captures a different, non-overlapping part of the total risk.

What is microstructure noise in intraday data?

It is the distortion in high-frequency prices from bid-ask bounce, discrete tick sizes and uneven liquidity, which makes returns look more volatile than the true underlying process. At fine bucket sizes it can substantially overstate an intraday volatility estimate unless it is explicitly corrected for.

How do execution desks use intraday volatility?

They schedule large orders around the U-shape, trading less into the violent open and more into the deeper, calmer midday and late-session liquidity, because trading when volatility and spreads are widest maximises market impact. Volume- and time-weighted strategies are built directly on the empirical shape of the session.

Why does the midday trough matter for risk?

Because the calm at the bottom of the U makes it cheapest and most comfortable to build a large position, which is then carried into the loud close and across the overnight gap. The quiet is a property of one bucket, not of the risk being taken, and treating it as safety is a common and expensive error.

Is intraday volatility forward-looking or backward-looking?

It is backward-looking. Like realised volatility, it is measured from prices that have already traded within the session, describing dispersion that has already happened. It is not a forecast, though its stable U-shape means the pattern of the day is fairly predictable even if the level is not.

Does the U-shape appear in every market?

It appears in essentially every exchange-traded market with a fixed open and close, because the mechanics are universal: information accumulates overnight and reprices at the open, and position-squaring and benchmark trading cluster at the close. The exact heights of the two arms vary by market, but the shape is remarkably consistent.

How does option gamma drive expiry-day intraday volatility?

Near expiry, a small move in NIFTY produces a large change in an at-the-money option's delta — that is high gamma — so hedgers must buy or sell the underlying aggressively to stay neutral. Their hedging flow itself moves the index, creating a feedback loop that makes expiry-afternoon intraday volatility spike or pin unpredictably.

Why is close-to-close volatility still used if it hides so much?

Because it is simple, needs only one price a day, and is the historical convention that everything else is compared against. It is adequate for many long-horizon purposes; its blindness to intraday shape and its blending of gap and session risk only bite when you have to act inside the day or size the overnight tail.

Voice search & related questions

Natural-language questions people ask about intraday volatility.

Why is the first few minutes of trading so wild?

Because the market was shut overnight while the world kept moving — US markets, currencies, company news — and at 9:15 the price has to catch up on all of it at once. That repricing is compressed into the first bucket, which runs about three times as volatile as a quiet midday one.

Why does volatility pick up again near the close?

Because traders are squaring positions they do not want to hold overnight, putting on their final directional views, and trading around the closing price, which sets settlements and benchmarks. The close is busy for position-management reasons, which is a different cause from the open's overnight-news repricing.

If two days had the same move, why do they feel so different?

Because close-to-close volatility only sees one number a day. One day might drift smoothly while another opens violently, reverses by lunch, and races into the close — identical close-to-close, completely different to trade. The intraday shape is exactly what that single daily number throws away.

Can my stop protect me overnight?

No. A stop only works while the market is open, so it cannot help against an overnight gap, where the price jumps from one close to the next open with no chance to trade in between. Gap risk is a separate, fatter-tailed danger that has to be sized on its own, not managed with a stop.

Why is expiry day so crazy on NIFTY weeklies?

Because as the options approach expiry their gamma explodes, so hedgers have to trade the index harder and harder to stay neutral, and that hedging pushes the index around. It can pin to the strike with the most open interest or snap violently when the pin breaks — a regime that looks nothing like a normal session.

Should I trust how calm the market feels at lunchtime?

Be careful. The midday quiet is the bottom of the U-shape, not the risk of the day. If you build a big position because it feels calm, you carry it into a loud close and across an overnight gap no stop can protect against. The calm is one bucket, not the whole picture.

How do the pros measure volatility better than close-to-close?

They use range-based estimators like Parkinson and Garman–Klass, which bring in the day's high, low and open, and they aggregate high-frequency intraday returns into a realised-volatility figure. Both use information the single closing price discards, so they converge on true volatility far faster from the same number of days.

Sources & references

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