Market Regimes Intermediate Oscillation around a centre Backward-looking (named after the fact)

Range-bound Markets

A range is only visible once it has held, and it stops being a range at the exact moment you have most confidence in it.

Quick answer: Range-bound markets are regimes of oscillation around a centre, where daily volatility can stay high while the multi-month return collapses toward zero — because the moves alternate in sign and cancel, the mirror image of a trend.

In simple words

A range-bound market oscillates around a centre: it moves up, then back down, then up again, without going anywhere on net. Suppose NIFTY spends months swinging between roughly 23,500 and 24,500 around a 24,000 centre, with daily moves of about 0.9% — a respectable 14% annualised. The days are not small, yet the 200-day return is close to zero, because the up-moves and down-moves alternate in sign and cancel. This is the mirror image of a trend: in a trend, small moves compound into a large return; in a range, large-ish moves cancel into no return at all.

The consequence for measuring risk is that realised volatility computed over a long window collapses while short-window volatility stays respectable. If you measure NIFTY's dispersion day-by-day it looks like a 14% market; if you measure the 200-day return it looks almost flat. The moves are alternating in sign — negative return autocorrelation — and that cancellation makes annualised daily volatility overstate the multi-month dispersion. It is the exact opposite of the trend, where the same annualising understates it. Same tool, same market, opposite error, depending only on the sign of the autocorrelation.

Not to be confused with: A low volatility market. A range can have high daily volatility and still go nowhere, because 'nowhere' comes from moves cancelling, not from moves being small. A calm market and a range-bound market can look identical on a long-window realised-volatility reading — both near zero net — while being completely different day to day, which is why the range fools traders into selling premium as though it were calm.

How respectable daily moves net to nothing

Busy days, no destination

A simulated index oscillating between 23,500 and 24,500 around a 24,000 centre with daily moves of about 0.9%.

The picture proves that daily size and net travel are independent in the other direction. Every bar is respectable, yet the path returns to where it started — so annualised daily volatility overstates how far this market actually goes. The range looks like skill to sell right up until the bar that leaves the box.

Professional explanation

High daily volatility, zero long-window return

The range-bound regime is the mirror image of the trend, and the arithmetic mirrors too. In a trend, small same-signed moves compound into a large return; in a range, respectable moves alternate in sign and cancel, so the 200-day return can be essentially zero while daily volatility stays high. The two facts — busy days and no net travel — are not in tension, because daily volatility measures the size of each day's move and the long-window return measures where the moves added up to, and cancellation drives a wedge between them. A trader who looks only at the daily figure sees a 14% market; a trader who looks only at the multi-month return sees a flat one. Both are correct, and the gap between them is the whole character of the regime.

Negative autocorrelation makes annualising overstate dispersion

The mechanism is negative return autocorrelation: in a range, an up-day tends to be followed by a down-day, because the move that pushed price toward the edge of the range is met by a move back toward the centre. This is mean reversion in price, and it does the opposite of what a trend's positive autocorrelation does to the √t rule. When returns are negatively autocorrelated, multi-period variance grows more slowly than the sum of one-period variances, so annualised daily volatility overstates the true multi-month dispersion. The 14% you compute from daily moves is an over-estimate of how far the market will actually travel over months — the opposite error to the trend, produced by the opposite sign of the same autocorrelation. This is why a range-selling strategy can look statistically favourable: the market genuinely moves less, over the horizons that matter to the position, than the daily volatility suggests.

The regime that builds a reputation and then destroys it

This is the regime in which iron condors and short strangles earn their reputation, and the regime whose end destroys them — and that sentence is the one a marketing department would cut. While the range holds, selling the edges works with almost monotonous reliability: price approaches a boundary, mean reversion pulls it back, the options expire worthless, and the seller collects. Every successful cycle deepens the conviction and, usually, the size. Then the range ends. The breakout that terminates a range arrives without warning from inside the range — there is no external signal, because the thing that defines a range is that price has not yet left it, and the session that leaves is indistinguishable from every session that reversed until the moment it does not reverse. A range is only visible once it has held, and it stops being a range at the exact moment you have most confidence in it, which is also the moment your position is largest.

Mean reversion in price is not mean reversion in volatility, and the variance ratio again

Two claims are constantly confused here, and they are not the same. Mean reversion in price — the idea that the index itself returns to a centre — is the assumption a range-seller is implicitly making, and its empirical support is weak: over most horizons index prices are close to a random walk, and clean, tradeable price mean reversion is far rarer than the people selling ranges assume. Mean reversion in volatility — the idea that the volatility level returns to a long-run average — is a different and far better-supported claim, because volatility genuinely clusters and reverts. Selling a range leans on the weak claim while borrowing credibility from the strong one. The formal diagnostic is the variance ratio test once more, now with the opposite sign: a variance ratio below 1 signals the negative autocorrelation of a range, just as a ratio above 1 signalled a trend. It tells you the recent window mean-reverted in price — it does not tell you the range will hold, and it cannot see the breakout coming, because the breakout comes from inside.

Formula

The variance ratio test, range-bound case

VR(q) = Var(r_q) / ( q × Var(r_1) ) · range ⇒ VR(q) under 1

The same statistic as for a trend, opposite sign. Negative return autocorrelation makes multi-period variance grow slower than q times the one-period variance, so VR(q) falls under 1 — the same statement as 'annualised daily volatility overstates a range's multi-month dispersion'. For first-order autocorrelation ρ₁, VR(2) ≈ 1 + ρ₁; a mild ρ₁ = −0.15 gives VR(2) ≈ 0.85.

VR(q)The variance ratio at horizon q. Equals 1 under independence, falls under 1 for a range (mean reversion), rises above 1 for a trend.
Var(r_q)Variance of the q-period (multi-day) return of the series.
Var(r_1)Variance of the one-period (daily) return of the series.
qThe aggregation horizon in periods — e.g. q = 20 tests monthly against daily returns.
ρ₁First-order autocorrelation of returns — negative in a range, where an up-day tends to be followed by a down-day.

Why annualising overstates a range

σ_multi-month = σ_daily × √q × √(VR(q))

With VR(q) under 1, the true multi-month volatility is smaller than the naive √q scaling of daily volatility — the factor √(VR(q)) is under 1 and pulls it down. This is the exact reason range-selling looks statistically favourable: the market travels less, over the position's horizon, than the daily number implies. It says nothing about the breakout that ends the range.

How to size a range without being destroyed by its end

Run the variance ratio test for a value under 1. Divide the variance of q-period returns by q times the variance of daily returns; a reading below 1 is the fingerprint of the negative autocorrelation that defines a range.
Do not confuse the two mean reversions. Selling a range assumes price mean-reverts, which the evidence supports weakly; do not borrow the credibility of volatility mean reversion, which is a different and better-supported claim, to justify it.
Size for the breakout, not the range. Because the session that ends a range is indistinguishable from every session that held until it does not, size the position so that a break through the boundary is survivable, not just the oscillation inside it.
Remember the position is largest when the regime is most confident. Successful cycles deepen conviction and size, so write down a maximum size before the range's reliability tempts you to raise it after each win.
Watch that the range is retrospective. A variance ratio below 1 tells you the recent window mean-reverted; it does not promise the range holds, so treat every boundary touch as a possible breakout rather than a guaranteed reversal.
Do not read a flat long-window volatility as calm. A range and a low volatility market both net near zero over long windows, but the range has respectable daily moves and a live breakout risk that the calm does not — check the daily figure and the variance ratio, not just the long-window number.

Practical example

NIFTY worked example

NIFTY oscillates between 23,500 and 24,500 around a 24,000 centre for several months, with a daily standard deviation of about 0.9% — which annualises to 0.9 × √252 ≈ 14.3%, a perfectly respectable figure. Yet measure the 200-day return and it is close to zero, because every push toward 24,500 is met by a push back toward 23,500. A trader sells an iron condor with short strikes near the range edges and collects, expiry after expiry, as price reverts each time it nears a boundary. Now interpret the two numbers together. The 14% daily volatility says this is a lively market; the near-zero 200-day return says it goes nowhere; and the variance ratio, printing well below 1, formally reconciles them by measuring the negative autocorrelation. The condor works precisely because the multi-month dispersion is smaller than the daily volatility implies — until the session that does not revert, which is drawn from inside the same range and looks identical to all the sessions that did, right up to the close that leaves the box.

BANKNIFTY worked example

BANKNIFTY sharpens the breakout half of the lesson. Suppose BANKNIFTY ranges between 50,000 and 54,000 around a 52,000 centre, and a trader sells strangles at the edges through several successful cycles, increasing size each time the range holds because the strategy 'keeps working'. Lot size is 30, so a strangle that has been collecting a few hundred points of premium per cycle is exposed to a breakout that, in a single stressed session, can move BANKNIFTY 850 points or more — 52,000 × 0.26 ÷ √252 at a stressed volatility. The breakout does not announce itself; it is the session that fails to revert, indistinguishable from the dozens that did until it closes outside 54,000. The NIFTY example shows a range reconciled by the variance ratio; the BANKNIFTY example shows why the reputation the range built is destroyed by its end — the same edges that paid reliably for months hand back the accumulated premium, and more, in the one cycle that breaks.

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. A range-bound market is where 'sell the edges, it keeps working' is most seductive and most dangerous, because the strategy's reliability is real right up until the breakout that ends it. The negative autocorrelation that makes annualised daily volatility overstate the range's dispersion is genuine, so the trade looks statistically favourable — but the breakout arrives without warning from inside the range, and it arrives when the position is largest and the conviction deepest. Short-volatility positions carry theoretically unlimited loss, and a range hands back its accumulated premium, and more, in the single cycle it fails to hold.

Advantages & limitations

What it is good for

It genuinely reduces multi-month dispersion. The negative autocorrelation is real, so annualised daily volatility overstates how far the market travels over the position's horizon — the range-seller is not imagining the favourable arithmetic while the regime holds.
It is diagnosable with the same variance ratio test as a trend. A reading below 1 objectively flags the mean-reverting behaviour, so the regime can be identified from price data without any option-pricing assumption.
It makes defined-risk premium structures legible. Iron condors and other bounded-loss structures have a clear rationale in a range, because the boundaries give a natural place to sell and a defined worst case if the range breaks.
It cleanly separates the two mean-reversion claims. A range is the environment that forces a trader to distinguish weak price mean reversion from well-supported volatility mean reversion, which is a distinction worth learning on.
It provides an explicit, testable exit condition. The range's boundaries define exactly where the thesis is wrong, so unlike many strategies the range-seller has an unambiguous line — the breakout — at which the position should be reconsidered.

Where it breaks down

The favourable arithmetic evaporates at the breakout. The overstatement of dispersion holds only while the range holds; the session that ends it is not covered by the statistic, and it is the session that matters.
The variance ratio is backward-looking. A reading below 1 describes the window it was computed over and cannot promise the range continues — the range, like every regime here, is named after the fact.
It relies on the weaker of the two mean reversions. Range-selling assumes price mean-reverts, which the evidence supports far less than volatility mean reversion, so the strategy's premise is shakier than its track record during the range suggests.
It masquerades as a low volatility market. A range and a calm market both net near zero over long windows, so a long-window reading alone cannot tell them apart, and a trader who sells a range as though it were calm mis-prices the daily risk and the breakout.
The breakout is unforecastable from inside. Because a range is defined by price not yet having left it, there is no in-range signal that distinguishes the session that breaks from the sessions that reverted — the diagnosis is structurally blind to its own end.
Conviction and size peak at the worst moment. The regime rewards adding size after each successful cycle, so the position is systematically largest exactly when the breakout risk it ignores is closest to being realised.

Common mistakes

Reading a flat 200-day return as a calm market. A range has respectable daily moves and a live breakout risk that a genuinely calm market does not, so selling a range as though it were low volatility mis-sizes both the daily risk and the tail.
Confusing mean reversion in price with mean reversion in volatility. Range-selling leans on the weakly-supported price claim while borrowing credibility from the well-supported volatility claim — a trader who does not separate them over-trusts the strategy's premise.
Adding size after every successful cycle. The range rewards this until it does not, so scaling up on the strategy's reliability makes the position largest at the exact moment the breakout that ends the range is closest.
Treating each boundary touch as a guaranteed reversal. The session that breaks the range looks identical to the sessions that reverted, so assuming the edge will hold again is assuming away the one outcome that destroys the position.
Ignoring the variance ratio and sizing off daily volatility alone. Without the ratio, a range and a trend with the same daily volatility are indistinguishable, and selling the edges of what is actually the early stage of a trend is a fast way to be run over.
Believing the breakout will announce itself. There is no in-range signal for the break, because a range is defined by price not having left it — waiting for a warning that cannot exist means the position is still full-size when the box is finally left.

Professional usage

Statistical arbitrage and mean-reversion desks are built to harvest exactly the negative autocorrelation a range produces, but they do it with a discipline retail range-sellers usually lack: they size to the breakout, not to the range. A stat-arb book treats the variance ratio below 1 as a signal that the recent window mean-reverted and simultaneously as a warning that the signal is retrospective, so it caps the loss on any single name at a level that survives the regime's end rather than optimising the income while it holds. Volatility desks, meanwhile, use the range to separate the two mean reversions cleanly — they will trade volatility mean reversion, which is well-supported, far more confidently than price mean reversion, which is not, and they price the difference into how much they are willing to sell at a range's edge.

Risk managers watch the variance ratio drifting below 1 with the same suspicion they watch it drifting above 1, because both are statements about the recent past that a position can over-trust. A book that has been quietly accumulating short-volatility exposure at a range's edges is, from a risk manager's chair, a position whose profit-and-loss is small-positive most cycles and catastrophically negative on the breakout — the same asymmetry as any short-volatility book, now dressed in the respectability of a reliable range. The professional response is to stress the book against a breakout to a stressed or crisis regime rather than against the range continuing, because the range continuing is the outcome the position already assumes, and the outcome that assumption ignores is the only one that hurts.

Key takeaways

A range-bound market oscillates around a centre so that daily volatility stays high while the multi-month return collapses toward zero — the mirror image of a trend, because the moves alternate in sign and cancel.
Negative return autocorrelation makes annualised daily volatility overstate multi-month dispersion, the opposite error to a trend, produced by the opposite sign of the same autocorrelation.
The variance ratio test detects it with a reading below 1 — the same statistic that reads above 1 for a trend — telling you the recent window mean-reverted in price.
Mean reversion in price, which range-selling assumes, is a weaker and far less supported claim than mean reversion in volatility, which genuinely clusters and reverts; the two are constantly and wrongly treated as one.
This is the regime that builds a reputation for iron condors and short strangles and then destroys it, because the breakout that ends the range arrives without warning from inside it, when the position is largest.

Learn that a range and a trend are the same tool read with opposite signs — the variance ratio above 1 for one and below 1 for the other — and the range-bound market becomes legible rather than treacherous. The negative autocorrelation is real and the favourable arithmetic is real, which is exactly what makes the regime dangerous: the strategy works, reliably, and every success deepens the conviction and the size, so the position is largest precisely when the breakout that ignores all of it is closest. A range is only visible once it has held, and it stops being a range at the exact moment you have most confidence in it. Size for the session that leaves the box, not the dozens that returned to the centre, because the two are indistinguishable until the close.

Frequently asked questions

What is a range-bound market?

A range-bound market is a regime of oscillation around a centre — the index moves up, then back down, without going anywhere on net. Daily volatility can stay high while the multi-month return collapses toward zero, because the moves alternate in sign and cancel. It is the mirror image of a trend, where small moves compound instead of cancelling.

How can daily volatility be high while the market goes nowhere?

Because daily volatility measures the size of each day's move and the long-window return measures where the moves added up to, and in a range they cancel. Respectable up-days and down-days alternate, so the 200-day return can be near zero even though the daily standard deviation is a lively 14% — the cancellation drives a wedge between the two.

What is negative return autocorrelation?

It is the tendency, in a range, for an up-day to be followed by a down-day — the move toward a boundary met by a move back toward the centre. It is mean reversion in price expressed day-to-day, and it makes multi-period variance grow more slowly than the sum of one-period variances, the opposite of a trend.

Why does annualising daily volatility overstate a range?

Because the √t scaling assumes independent returns, and a range violates that with negative autocorrelation. When up-days are followed by down-days, multi-period variance grows slower than √t predicts, so annualised daily volatility over-estimates how far the market actually travels over months — the exact opposite of the trend's under-estimate.

How does the variance ratio test detect a range?

By printing below 1. The variance ratio is the variance of q-period returns divided by q times the variance of daily returns; negative autocorrelation makes multi-period variance small, so the ratio falls under 1. It is the same statistic that reads above 1 for a trend — same tool, opposite sign.

Why do iron condors work in a range-bound market?

Because the negative autocorrelation is real, so price tends to revert from the boundaries and the sold options expire worthless cycle after cycle. The multi-month dispersion is genuinely smaller than the daily volatility implies, which makes the trade look favourable — while the range holds. The breakout that ends the range is exactly what the structure does not survive well.

Why do range-selling strategies eventually blow up?

Because the breakout that terminates a range arrives without warning from inside it, when the position is largest. Every successful cycle deepens conviction and size, so the seller is most exposed at the moment the range is closest to breaking — and the single cycle that fails to revert hands back the accumulated premium and more.

Is mean reversion in price the same as mean reversion in volatility?

No, and confusing them is a core error of range-selling. Mean reversion in price — the index returning to a centre — is weakly supported; over most horizons prices are close to a random walk. Mean reversion in volatility — the volatility level returning to average — is well-supported, because volatility clusters and reverts. Range-selling leans on the weak claim.

How is a range different from a low volatility market?

A range has respectable daily moves that cancel; a low volatility market has small daily moves. Both net near zero over long windows, so a long-window reading alone cannot tell them apart — but the range carries a live breakout risk and a higher daily risk that a genuinely calm market does not, which is why selling a range as though it were calm mis-sizes the trade.

Can I predict when a range will break?

No, not from inside the range. A range is defined by price not yet having left it, so there is no in-range signal that distinguishes the session that breaks from the sessions that reverted — they look identical until the close that leaves the box. The variance ratio confirms the range existed; it cannot see its end.

What does a variance ratio below 1 tell me?

That the series mean-reverted over the window measured — multi-period variance is smaller than q times the one-period variance, the fingerprint of negative autocorrelation. Practically, it warns that annualised daily volatility overstates the multi-month dispersion, which is why range-selling looks statistically favourable while the ratio stays below 1.

Why is the breakout so dangerous for a range-seller?

Because it arrives when the position is largest and the conviction deepest, and it does not announce itself. The same edges that paid reliably for months are where the short options sit, so a single session that closes outside the range turns a string of small wins into a loss that can exceed all of them combined.

How do I tell a range from a trend if daily volatility is the same?

By the variance ratio, since the volatility level cannot distinguish them. A range prints below 1 (negative autocorrelation), a trend prints above 1 (positive autocorrelation). Two markets with identical daily volatility can be a range and a trend, and selling the edges of what is actually an early trend is a fast way to be run over.

Does a range-bound market have a lot of risk?

It has more than its calm appearance suggests. The daily moves are respectable, the breakout risk is live and unforecastable, and short-volatility positions sold at the edges carry theoretically unlimited loss. The favourable multi-month arithmetic is real while the range holds and irrelevant the session it breaks.

What is the biggest misconception about range-bound markets?

That a market going nowhere is a market with little risk. 'Nowhere' comes from moves cancelling, not from moves being small, so a range can be a lively 14% market day-to-day with a breakout waiting — the flat long-window return hides both the daily risk and the tail.

How should I size an iron condor in a range?

Size it to survive the breakout, not to optimise the income while the range holds. Because the session that ends the range is indistinguishable from the ones that reverted, and because conviction tempts you to add after each win, write down a maximum size in advance and treat every boundary touch as a possible break rather than a guaranteed reversal.

Why is price mean reversion weaker evidence than it seems?

Because over most horizons index prices behave close to a random walk, and clean, tradeable price mean reversion is rarer than range-sellers assume. A range that has held is retrospective evidence of past mean reversion; it is not proof the mechanism will pull price back from the next boundary touch, which is the assumption the trade actually needs.

How do statistical arbitrage desks trade a range?

They harvest the negative autocorrelation but size to the breakout rather than the range, capping the loss on any single position at a level that survives the regime's end. They treat a variance ratio below 1 as a retrospective signal and never as a promise, which is the discipline that separates them from a range-seller who scales up on reliability.

Does a flat market mean I can safely sell premium?

Not safely — the flatness can be a range with respectable daily moves and a live breakout, or a genuinely calm market, and only the daily volatility and the variance ratio tell you which. Selling premium into either carries theoretically unlimited loss, and a range specifically hands it back at the breakout. None of this is investment advice.

Why do range and trend use the same test with opposite signs?

Because both are departures from independence, in opposite directions. A trend is positive autocorrelation (variance ratio above 1); a range is negative autocorrelation (variance ratio below 1). The variance ratio measures how fast multi-period variance grows relative to daily variance, so it reads high for the trend that compounds and low for the range that cancels.

Voice search & related questions

Natural-language questions people ask about range-bound markets.

How can a busy market end up going nowhere?

Because the moves cancel instead of compounding. In a range an up-day tends to be followed by a down-day, so respectable daily moves add up to almost no net travel — daily volatility stays high while the 200-day return collapses toward zero, the exact mirror of a trend.

Why does selling the edges of a range work so well until it doesn't?

Because the negative autocorrelation is real, so price keeps reverting from the boundaries and the sold options keep expiring worthless — until the session that fails to revert. The strategy's reliability deepens your conviction and size, so you are largest exactly when the breakout that ignores all of it arrives.

Can I tell when the range is about to break?

No, not from inside it. A range is defined by price not having left it, so the session that breaks looks identical to every session that reverted right up to the close that leaves the box. The variance ratio can confirm the range existed, but it is structurally blind to the breakout.

Isn't a range-bound market just a calm market?

No — they only look alike over long windows, where both net near zero. A range has respectable daily moves and a live breakout risk that a genuinely calm market does not, so treating a range as calm and selling premium into it mis-sizes both the daily risk and the tail.

People say the market mean-reverts, so isn't range-selling safe?

Mean reversion in price is far weaker evidence than it sounds — over most horizons prices are close to a random walk. What is well-supported is mean reversion in volatility, which is a different claim. Range-selling leans on the weak one while borrowing the credibility of the strong one, and no version of it is risk-free.

Why does my condor keep winning and then give it all back?

Because a range hands back its accumulated premium, and more, in the one cycle it fails to hold. The wins are the range reverting from its edges; the giveback is the breakout, which arrives without warning from inside the range when your position is largest and your confidence highest.

Sources & references

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