Guide · Regime and risk · 2026-04-25

Regime detection in Indian markets: the methods and their limits

The short answer

Regime detection classifies the market's current character, trending or ranging, calm or stressed, so you can tell which kind of strategy has a fair chance of working. In Indian equities you read it from three families of signal: trend, through moving-average slope and stacking, swing structure and the ADX; volatility, through India VIX as the forward gauge plus realised measures like ATR; and breadth, through advance-decline and the share of stocks above a moving average. The hard truth is that every one of these lags: a regime read is a probability tilt and a risk governor, not a forecast of direction.

The companion piece, why filtering trades by regime matters, argues the case for conditioning a strategy on the market's state. This page is the other half: the concrete methods for actually reading the current regime, and the failure modes that make regime detection harder than a neat two-by-two grid suggests. A strategy's edge is regime-conditional. A trend-following system dies the death of a thousand cuts in a range, and a mean-reversion system can be run over by a genuine trend. So the practical question is not whether regime matters but how you read it without fooling yourself, given that the tools confirm a regime only after it has begun.

Trend regime: slope, structure and strength

A trend regime is one where price is going somewhere and staying there. Three readings triangulate it, and they are strongest when they agree.

The first is moving-average slope and stacking. In a clean uptrend a faster average sits above a slower one and both slope up; in a downtrend the order inverts and both slope down; in a range the averages flatten, cross repeatedly and offer no information. A common macro filter is a single long average, often the 200-day, used as a coarse switch: price and slope above it lean the book toward long trend systems, below it toward defence. The tool is blunt on purpose, and its bluntness is what makes it slow.

The second is swing structure, the most model-free reading of all. An uptrend prints higher highs and higher lows; a downtrend prints lower highs and lower lows; a range prints overlapping swings that keep returning to the same band. Structure needs no parameter, but it is only legible after the swings have formed, so like everything here it is a lagging read.

The third is the Average Directional Index (ADX), introduced by J. Welles Wilder in 1978. Its single most useful property is that it measures trend strength irrespective of direction: a strong uptrend and a strong downtrend can both post a high ADX. Wilder's convention reads roughly above 25 as a trend present and below 20 as no trend, with an ambiguous grey zone between. Because the ADX is built from several layers of smoothing, it lags noticeably and often confirms a trend only after a good part of the move has passed. That lag is the price of its steadiness.

The regime map organises the decision, it does not draw exact lines. The horizontal axis is trend strength, from range to trend; the vertical axis is volatility, from calm to stressed. Each quadrant points to the approach with the better odds in that state, and rising volatility argues for smaller size everywhere. The grid is illustrative: real markets slide between quadrants, and the boundaries are fuzzy.

Volatility regime: India VIX, ATR and clustering

Volatility is the second axis, and it is often the one that decides your size rather than your direction. The forward gauge in Indian markets is India VIX. NSE computes it from the order book of Nifty 50 options, using the methodology the Chicago Board Options Exchange pioneered for its VIX and licensed to NSE, adapted to the Nifty book with cubic-spline interpolation across strikes. It reads the best bid-ask quotes of out-of-the-money Nifty calls and puts across the near and next monthly expiries, and expresses the market's expected volatility of the Nifty over the next 30 days as an annualised percentage, refreshed roughly every fifteen seconds through the session. In plain terms, it is the price the market is putting on protection.

India VIX and the Nifty usually move in opposite directions, and the mechanism is demand for insurance, not superstition. A falling market makes holders reach for put options and makes writers demand more, premiums swell, and the VIX, read out of those premiums, jumps. A calm uptrend lets the urgency fade and premiums soften, and the VIX drifts down. The average correlation between the Nifty and India VIX has been about minus 0.41 on data since 2008, which makes it a strong tendency rather than a mechanical rule. For context, the index has spent most of its life in the low-to-mid teens, while its extreme readings came from crises: it spiked into the high 80s during the COVID-19 crash of March 2020, and on the reconstructed 2008 series reached the low 90s intraday. Those spikes show how far the gauge can travel when protection is scrambled for.

The realised counterpart is the ATR idea, the Average True Range that Wilder also introduced in 1978. ATR is a smoothed average of the true range of each bar, and its refinement is that it uses the true range rather than plain high minus low, so it captures the overnight gaps that a naive range would miss. Where VIX is forward and implied, ATR is backward and realised: it tells you how much price has actually been moving, which is why it is the standard input for volatility-scaled position sizing and stop distance. It lags, and it understates risk into scheduled binary events, where implied volatility can jump before any realised move.

Underpinning both is volatility clustering, the empirical fact that calm tends to follow calm and stress tends to follow stress. Benoit Mandelbrot observed in 1963 that large price changes tend to be followed by large changes of either sign and small by small, and that observation was later formalised in the ARCH model of Engle in 1982 and the GARCH model of Bollerslev in 1986. Clustering is what makes a volatility regime a usable idea at all: because volatility is persistent, today's reading carries information about tomorrow's, so a stressed regime is worth respecting until it visibly subsides rather than fading it on the first calm day.

The bands are a rough convention, not a rule. A low VIX marks a calm regime, a middling VIX nervousness, and a high VIX genuine stress, but the boundaries shift with the era. Read the level against the index's own recent range rather than an absolute line, and note the shape: fear spikes fast and eases slowly, which is volatility clustering made visible.

On the numbers. The band levels above and any VIX or ADX thresholds in this guide are rough conventions, not fixed lines, and the whole distribution shifts with the regime. Treat any hard cut-off you see quoted, here or elsewhere, as illustrative, and verify the current level and its context before using it.

Breadth: the participation behind the index

Trend and volatility describe the index. Breadth describes the army behind it, and a move led by a shrinking number of names is a weaker regime than the headline suggests. Three gauges are standard.

The advance-decline line cumulates the number of advancing stocks minus declining stocks each day into a running total. A rising line says participation is broad and the trend is healthy; a falling line says more stocks are sinking than rising beneath a flat or rising index. The percentage of stocks above a moving average, often the 50-day or 200-day, is the most intuitive breadth reading: a high share signals broad participation, a low share signals a narrow, fragile advance. And new highs against new lows tracks how many names are actually making fresh extremes, expansion confirming a trend and contraction warning of fatigue.

The reading that matters most is breadth divergence: the index grinds to a new high while these gauges roll over and make a lower high, meaning fewer and fewer stocks are carrying the move. Because participation usually deteriorates before the index itself takes damage, a divergence is an early warning that the trend regime is thinning from the inside. The catch is that divergences can run for a long time before they resolve, so a divergence is a caution to tighten risk and demand more from new longs, not a timing signal to sell.

The index makes a higher high while breadth makes a lower high. Participation is draining out of the advance even as the headline number climbs. That gap is the warning: the trend regime is narrowing, and the odds of it stalling are rising. It is a reason to demand more from new longs and to protect gains, not a precise sell trigger.

Combining the signals into a regime read

No single gauge defines a regime; a read comes from making the families agree. Trend answers is price going somewhere, volatility answers how violent is the tape, and breadth answers how many names are behind it. The most confident states are the ones where they line up: a rising, well-stacked average with an ADX above its trend line, a calm-to-moderate VIX and broad breadth is a healthy trend regime; a flat, tangled average with a low ADX and a middling VIX is a range. The uncomfortable states, and the ones that lose money, are the disagreements: a fresh index high on narrowing breadth, or a strong-looking trend as the VIX climbs into stress.

Once you have a read, its job is to set your posture, not to predict the next tick. It decides which strategy is even allowed to trade, and it scales your size: full size in a quiet trend, smaller and more patient in a volatile one, and out of the way when the volatility regime is deep in stress. Reading that state, and letting it govern which idea you deploy and how large, is upstream of any single setup, and that upstream discipline is exactly what the method we teach is built around. The table below collects the toolkit in one place.

The regime-detection toolkit: signal, what it measures, the regime it flags, and its main limitation
Signal	What it measures	Regime it flags	Its lag or limitation
MA slope and stacking	Direction and order of fast versus slow averages	Trend up, trend down or flat range	Smoothed, so slow; whipsaws when averages tangle in a range
Swing structure	Higher-highs and higher-lows versus overlapping swings	Trend versus range	Only legible after swings form; discretionary to read
ADX	Trend strength, ignoring direction	Trend present or absent	Heavily smoothed; confirms late; grey zone 20 to 25
India VIX	Option-implied 30-day expected Nifty volatility	Calm, elevated or stressed	Reflects fear, not direction; no fixed band
ATR	Realised average true range per bar	Expanding or contracting volatility	Backward-looking; understates risk into events
Advance-decline line	Cumulative advancers minus decliners	Broad versus narrow participation	Divergences can run long before resolving
Percent above a MA	Share of stocks above a key moving average	Healthy versus fragile breadth	A level, not a trigger; needs context

The regime-to-approach map, in words. Illustrative postures, not instructions
Regime	Typical trend read	Typical volatility read	Approach with better odds
Quiet trend	Stacked averages, ADX in trend zone	Calm to moderate VIX	Trend following, full size, ride pullbacks
Volatile trend	Trending but choppy	Elevated VIX, wide ATR	Trend following, reduced size, wider stops
Quiet range	Flat, tangled averages, low ADX	Calm VIX	Mean reversion, fade the band edges
Volatile range	Directionless	Elevated VIX, wide swings	Smaller, patient reversion; fewer trades
Deep stress	Direction unreliable	VIX in the stressed band	Preserve capital; stand aside until it subsides

The honest limits: lag, whipsaw and overfitting

This is the part most treatments skip, and it is the part that decides whether regime detection helps you or flatters you. Three limits are structural.

First, regime detection is inherently lagging. Nearly every tool here is built from smoothed history: moving averages, ADX and ATR all average past bars, so they can only confirm a regime after it has begun, and the smoother the tool the greater the delay. India VIX is more immediate because it reads live option premiums, but it reports the price of fear, not the direction of price. The honest consequence is that you generally recognise a trend once it is under way and a stress regime once volatility has already expanded. You are reading the weather, not forecasting it.

Second, regime reads whipsaw. When a signal hovers near its boundary, an ADX oscillating around 20 to 25, a moving-average slope flattening, a VIX loitering on a band edge, the label flips back and forth and manufactures a run of false switches. That is worst precisely in the sideways, indecisive conditions where trend tools are already weakest, so the moments you most want a clean read are the moments the read is least stable. The standard defence is hysteresis: require a signal to move well past a threshold before you flip the label, and well back before you flip it again, accepting more lag in exchange for fewer false alarms.

Third, over-tuned regime rules are just another form of overfitting. If you keep adjusting thresholds until the rules would have classified past data perfectly, you have described history rather than discovered structure, and the fitted rule tends to fall apart out of sample. A regime read earns trust when it rests on a few sturdy, economically sensible signals rather than a lattice of hand-picked cut-offs. Simpler and slightly wrong tends to survive; precise and overfit tends not to.

What a regime read is, and is not. A regime read is a probability tilt and a risk governor. It shifts the odds toward the right kind of strategy and it scales your exposure, and that is genuinely valuable. It is not a predictor: it will not tell you the top, the bottom or the next move, it will lag the turns, and it will occasionally flip on noise. Treat it as a way to avoid fighting the tape, not as a crystal ball, and size for the fact that it is sometimes wrong.

Where a regime read fits

A regime read sits above the individual trade, in the layer that decides which game you are playing before you look for a setup. Read plainly, it keeps you from running a mean-reversion idea into a freight-train trend, or a breakout idea into chop that grinds it down, and it makes you smaller or absent when volatility is stressed. It cannot manufacture an edge, and it cannot see the future. What it can do is stop you deploying the wrong tool in the wrong weather, which over a long enough run is worth more than any single well-timed trade. For the argument on why to filter at all, and how a regime filter fits a working process, see the companion page linked below.

Read the market's state before you trade it.

Bharath Shiksha is a 30-volume curriculum across 6 stages, from chart reading (Stage 1 at ₹14,999) through capital raising (Stage 6 at ₹59,999), or the full bundle at ₹1,49,999. The upstream skill of reading regime and governing risk runs through it, well before any single setup. Start with the free diagnostic to see where you stand.

Take the free diagnostic →

Frequently asked questions

What is regime detection in trading?

Regime detection is the practice of classifying the market's current character so you know which kind of strategy has a fair chance of working. The usual axes are trend versus range and calm versus stressed volatility. It matters because a strategy's edge is conditional on regime: a trend-following system bleeds by small cuts in a sideways market, and a mean-reversion system can blow up when a real trend runs against it. Detecting the regime is a probability tilt and a risk governor, not a forecast of direction.

How do you tell if the Indian market is trending or ranging?

Three readings agree in a trend and disagree in a range. First, the slope and stacking of moving averages: in an uptrend a faster average sits above a slower one and both point up. Second, swing structure: a trend prints higher highs and higher lows, while a range prints overlapping swings that go nowhere. Third, the Average Directional Index, or ADX, which measures trend strength regardless of direction. J. Welles Wilder's convention reads roughly above 25 as a trend and below 20 as no trend, with a grey zone between. All three lag, so they confirm a move rather than call it early.

How is India VIX used to read the volatility regime?

India VIX is the market's forward volatility gauge. NSE computes it from the order book of Nifty 50 options using the methodology the Chicago Board Options Exchange pioneered for its VIX, adapted to the Nifty book, and it expresses the expected volatility of the Nifty over the next 30 days as an annualised percentage. A low reading signals a calm regime, a middling reading signals nervousness, and a high reading signals stress. There is no fixed line: the low-to-mid teens have been typical of calm, the twenties of nervousness, and the thirties and beyond of genuine stress, but treat any quoted cut-off as a rough convention and verify the current level.

Why do India VIX and the Nifty usually move in opposite directions?

The link runs through demand for protection. When the Nifty falls sharply, holders scramble for downside insurance, buyers reach for put options and sellers demand more to write them, so option premiums swell across strikes. India VIX is read straight out of those premiums, so it jumps. In a calm uptrend the urgency to hedge fades, writers compete and premiums soften, so the VIX drifts lower. The average correlation between the Nifty and India VIX has been about minus 0.41 on data since 2008, so the inverse relationship is a strong tendency, not a mechanical law.

What is market breadth and how does breadth divergence warn of a weakening regime?

Breadth measures how many stocks are participating, not just where the index sits. Common gauges are the advance-decline line, which cumulates advancing stocks minus declining ones, the percentage of stocks trading above a moving average, and the count of new highs against new lows. Breadth divergence is when the index makes a fresh high while these gauges make a lower high, meaning fewer and fewer names are carrying the move. Because participation usually deteriorates before the index itself does, a breadth divergence is an early warning that the trend regime is thinning, though the timing is imprecise.

What is the difference between India VIX and ATR as volatility measures?

India VIX is forward-looking and implied: it is derived from option prices and states what the market expects volatility to be over the next 30 days. ATR, the Average True Range that Wilder introduced in 1978, is backward-looking and realised: it is a smoothed average of how far price has actually travelled each bar, and it captures overnight gaps because it uses the true range rather than plain high minus low. VIX tells you what protection is being priced for the future, while ATR tells you how much the instrument has been moving. Both lag around scheduled events, where implied volatility can spike before realised volatility does.

Why is regime detection always lagging?

Every regime tool is built from smoothed history. Moving averages, ADX and ATR all average past bars, so they can only confirm a regime after it has begun, and the smoother the tool the greater the lag. India VIX is more immediate because it reads live option premiums, but it reports fear rather than direction. The consequence is that you generally recognise a trend once it is underway and a stress regime once volatility has already expanded. Regime detection reduces the odds of applying the wrong strategy; it does not let you step in front of the change.

Can over-tuning a regime filter cause problems?

Yes. Two failures recur. The first is whipsaw: when a signal such as a moving-average slope or an ADX level hovers near its boundary, it flips back and forth between regime labels and generates a run of false switches, exactly the sideways condition trend tools handle worst. The second is overfitting: if you tune thresholds until they would have classified past data perfectly, you have described history, not the future, and the fitted rule tends to fail out of sample. A robust regime read uses a few sturdy signals with hysteresis and treats the label as a probability tilt, not a precise switch.

Where the facts come from

Sources

India VIX methodology. NSE computes India VIX from the Nifty 50 options order book using the CBOE VIX methodology, licensed and adapted with cubic-spline interpolation, expressing expected 30-day Nifty volatility as an annualised percentage, refreshed through the session. nseindia.com
India VIX and Nifty relationship. The inverse tendency is documented, with the average Nifty to India VIX correlation reported at about minus 0.41 on data since 2008, and crisis peaks in the high 80s (March 2020) and low 90s intraday on the reconstructed 2008 series. motilaloswalamc.com
ADX, trend strength and lag. The Average Directional Index, developed by J. Welles Wilder in 1978, measures trend strength irrespective of direction; a strong trend is read above about 25 and no trend below about 20, with a grey zone between, and the indicator lags because of its smoothing. chartschool.stockcharts.com
ATR as a realised-volatility measure. The Average True Range, introduced by Wilder in his 1978 book, is a smoothed average of the true range that captures overnight gaps and serves as the standard input for volatility-scaled sizing and stops. en.wikipedia.org
Volatility clustering. Benoit Mandelbrot observed in 1963 that large changes tend to follow large and small to follow small; the effect was formalised in the ARCH model (Engle, 1982) and the GARCH model (Bollerslev, 1986). en.wikipedia.org
Market breadth and divergence. The advance-decline line and the percentage of stocks above a moving average gauge participation; a breadth divergence, an index high on a lower breadth high, warns that participation is narrowing before the index itself weakens. fidelity.com

Educational note. This guide explains how to read the market's regime and the limits of doing so. It is not a recommendation to trade or invest, and it is not investment advice. A regime read manages risk and tilts probability; it does not forecast. Bharath Shiksha is an educational publisher, not a SEBI-registered investment adviser or research analyst. All charts are illustrative and labelled as such.