Two fear gauges, and the gap between them

A volatility index is a price, not a forecast

India VIX and the US VIX both measure the same thing on different markets: how much movement option prices imply over the next 30 days, annualised and quoted as a percentage. A high reading means protection has become expensive. It says nothing about which way prices go.

That distinction is the whole article, so it is worth making concrete. An option's price is what somebody paid for the right to buy or sell later at a fixed price. Work backwards from that price through a pricing model and you can ask what level of future movement the buyer must have been assuming to pay it. That level is the implied volatility — how it is read off the chain strike by strike is the subject of option chain analysis. A volatility index does the same job across a whole strip of strikes and prints one number.

Now notice what is missing. A buyer of a put and a buyer of a call are both paying for movement, and both purchases push the index up. The index is symmetric by construction: it cannot distinguish fear of a fall from anticipation of a jump. Calling it a fear gauge is a convention, and a slightly misleading one — what it actually gauges is the price of not knowing.

The reason it has nevertheless kept company with falling markets is a fact about behaviour rather than about the formula. Most portfolios are long, so a fall is what hurts them, and demand for downside cover is the demand that spikes. That is why readings have risen in selloffs — not because the arithmetic points down, but because that is when people bid for insurance. Hold on to the difference. It is the reason a high reading is a statement about crowding into protection and not a statement about direction.

The same recipe, cooked in two different kitchens

The two gauges are not cousins. They are the same calculation applied to different underlyings — NSE licensed the methodology, which is why the outputs are directly comparable in units even when they are nowhere near comparable in meaning.

Three features they share, and each one matters later:

• Model-free. Neither one inverts Black–Scholes strike by strike. Both aggregate a strip of out-of-the-money option prices across the near and next expiry into a single variance figure, so no single strike's quirks dominate.
• Read off the order book, not off trades. Both are computed from standing bid and ask quotes rather than last traded prices. That is a deliberate choice: an illiquid strike that last traded an hour ago would otherwise inject a stale number into a live index.
• Thirty days, annualised. Both look 30 days ahead and both scale the answer up to a yearly figure, which is the convention that makes them look larger than the movement anyone actually expects tomorrow.

What differs is only the option book each reads. The US VIX reads S&P 500 options. India VIX reads Nifty 50 options. Everything that follows — every divergence, every argument in this article — comes from that one difference, because an option book prices the events its own market is facing.

What a reading of 14 actually says

Take an illustrative figure — not a quote, and not today's — and say the gauge reads 14. Here is how to turn that into something you can picture.

The 14 is an annualised one-standard-deviation figure: a measure of how far returns spread around their average, glossed properly in mutual fund risk measures. Volatility scales with the square root of time, so to bring an annual figure down to a shorter period you divide by the square root of the number of those periods in a year.

• One month: 14 ÷ √12, which is 14 ÷ 3.46 ≈ 4.0%.
• One day: 14 ÷ √252, which is 14 ÷ 15.9 ≈ 0.9%.

So an annualised 14 is the option market pricing daily moves of roughly 0.9%, and a one-month range of about 4% either side. Under a normal distribution you would expect the outcome inside one standard deviation roughly two times in three.

That last sentence contains the assumption worth distrusting. Equity returns are not normally distributed — the extremes arrive more often and land further out than the bell curve allows. So the one-in-three of outcomes falling outside the band are not spread evenly around it; a disproportionate share of them are the large ones. The arithmetic above is a way of picturing the number, not a probability you can bank.

The two gauges side by side

Where they are identical, and where they are not.

The bottom three rows are FNOTrader's design choices rather than properties of the indices, and the section on the model says why they are set that way.

Why they disagree — the calendar is inside the number

The single most useful thing to understand about India VIX is that it is not a domestic copy of the US VIX running a few hours behind. It prices a different set of scheduled events, and those events are dated.

An option expiring in three weeks has to be priced for everything that will happen in those three weeks. If a Union Budget, an RBI policy decision or an election counting day falls inside the window, the option seller demands more for taking the other side, because a single scheduled morning can move the index further than a normal fortnight. The reading rises because the calendar changed, not because sentiment did. The reverse is mechanical rather than a tendency: once the date falls outside the contract's remaining life, the event is no longer among the things that contract has to be priced for, and whatever premium was being charged to carry it has nothing left to price. What the book does with the rest of its uncertainty on that day is a separate question, and not one the calendar answers.

The US VIX carries its own calendar: policy meetings, inflation releases, the US election cycle. Those dates do not line up with India's, and neither book knows about the other's.

This is the named failure mode, and it catches experienced people: the calendar is inside the number. An option seller looks at a domestic gauge sitting well above its recent range, concludes that premium is rich, and sells into it. Sometimes that is a genuine dislocation. Sometimes the elevated reading is a scheduled event three weeks out that the seller has just agreed to carry, at a price they judged by comparing today's number with last month's — a comparison in which the event was invisible. Before treating any volatility level as expensive, check what is inside the 30 days it covers. The mechanics of what a seller is actually taking on are in straddle versus strangle.

There is a second, quieter source of divergence: the two indices sit on different underlyings, and Nifty and the S&P 500 do not move the same amount on an ordinary day. Their typical levels therefore differ for reasons that have nothing to do with stress. Which leads to the comparison people make most often, and should not.

Comparing the two levels directly is a category error

Both indices are quoted in the same units, so it is tempting to read one against the other. Keep the illustrative 14 from the last section, put an equally invented 18 next to it on the US gauge, and the inference writes itself: India must be the calmer market. Same units, different scales of normal.

The units are annualised percentage volatility in both cases. What differs is the distribution each number is drawn from. An index whose components move more on a routine day carries a higher baseline reading in quiet conditions and in stressed ones alike, and that baseline is a fact about the constituents, their sector mix and their liquidity — not a fact about fear.

So the comparison that carries information is each gauge against its own history, and then the two positions against each other. A gauge in the top decile of its own range and one in the middle of its own range is a real divergence. Two raw numbers three points apart is not. This is the same discipline the macro page applies to every tile: read a level against its own range rather than against a memory of it, which is why each tile opens its own history chart.

The trade-off in doing it properly is that you need the history to hand, and a percentile is slower to read than a number. It is also the difference between a comparison that means something and one that only looks like it does.

Reading the gap: four configurations

Once you stop reading either level on its own, the pair resolves into four states. Each one is a question about where the risk being priced lives.

1. Both quiet. Neither book is paying up for protection. Worth remembering that a low reading is a price, not a promise — protection being cheap is a statement about what the market is charging today, and cheap insurance is exactly what a market that has been calm for a while tends to offer.
2. US up, India flat. The stress is priced abroad and has not been priced here. That is a genuine piece of information, and it has at least two readings: the shock is one Indian earnings and Indian policy are insulated from, or the domestic book has not repriced yet. The page shows you the disagreement. It does not resolve it, and nothing in the mechanism says which side moves. What you can do is name the route by which it would arrive if it does — the dollar, global rates, energy, credit, foreign flows — which is what the macro signals map is for.
3. India up, US flat. The cause is local, and there is usually a date attached: a Budget, a policy decision, a counting day, a domestic credit event. This is the configuration where the calendar explanation should be your first hypothesis, because it is the commonest one and it is checkable in seconds.
4. Both up together. A global risk event that both books are pricing. Here the volatility pair is confirming rather than informing — it is saying the same thing credit spreads and the dollar are saying, and credit is where that class of event is conventionally watched for first, for the structural reason set out in credit spreads and where trouble shows up first. Whether it actually arrives there first on any given occasion is a claim about that occasion, not a property of the plumbing.

One discipline holds across all four. A gap is an observation about a particular day; a channel is a mechanism. "The two gauges have diverged" is something you can see. "Global stress has not been imported yet" is an interpretation you are adding, and it is only as good as the route you can name. Correlation between the two series over 30 or 90 sessions is the same kind of claim — a measurement of those sessions, which is why a coefficient that flips sign between windows is telling you about the window rather than about the relationship.

Why our model weights both, and weights them differently

Both are inputs to the composite score on the macro page, and the US gauge is weighted heavier because our model treats it as the import channel and the domestic gauge as local confirmation. The full formula and the reason the score measures agreement rather than severity belong to the macro signals pillar; what matters here is how the volatility pair is treated inside it, because the treatment encodes an argument.

In outline: each input's one-day percentage change is divided by a saturation scale, clamped to the range −1 to +1, signed for whether rising helps or hurts Indian equities, and the contributions are then averaged with weights. The US VIX carries a weight of 0.10 and India VIX 0.06, both signed negative. Every one of those numbers is FNOTrader's judgement about what has mattered to Indian equities, not a measured constant and not the output of a regression. Someone with a different reading would weight them differently and would not be wrong on the arithmetic.

Three things follow, and the third is the interesting one.

Neither gauge dominates. There are twenty weighted market tiles and they sum to 1.33 — not a tidy handful. The denominator is not that constant, though: it is the sum of the weights that actually reported, so an input whose feed failed drops out of both halves of the average, and foreign-flow data adds a further weight of its own on the days it publishes, which takes the all-present divisor to 1.45. So the two volatility tiles carry 0.16 between them against that 1.45 when every input reports — a little over a tenth of the score — a slightly larger share on a day the flow figure does not land, and larger still on a day several feeds fail. A volatility spike alone does not move the gauge far, and that is the design working rather than failing.

Both are signed negative. Rising means red on both tiles, because a dearer price of protection is a headwind for equities in our reading. Note that this is one of the few places on the macro page where the colour convention is not counter-intuitive — unlike USD/JPY, where a rising pair shows green because a weakening yen keeps the yen carry trade intact.

And the asymmetry, which is the point. The two tiles share the same saturation scale, so an identical one-day percentage move in each fills the same fraction of its ceiling — and after that, the weights decide. At 0.10 against 0.06, the same percentage move in the US gauge counts roughly one and two-thirds times as much toward the score as it does in the domestic one. Then the stress override sharpens it further: a one-day jump of 20% or more in the US VIX flips the page to Stress on its own, bypassing the average entirely, and a move of any size in India VIX does not.

Read together, those two choices are the import-channel reading made concrete. A global volatility shock is the class of event that can arrive through every other channel at once and with no entry in anyone's diary, so it gets the escape hatch. A domestic spike usually has a date attached and is already visible in the calendar, so it does not. That is an argument, not a measurement, and a reader who thinks domestic events deserve the override is disagreeing with a design choice rather than finding a bug. It is published for exactly that reason.

Why a high reading is not automatically an opportunity

The conventional reading among option sellers is that implied volatility usually prints above the volatility that subsequently arrives, which is what makes systematic selling look attractive on paper.

Treat that as what it is. It is an empirical claim about a market and a period, not a mechanism — and unlike a mechanism it can stop being true without anything announcing it. Whether it holds on Nifty options over any particular stretch is a question to be answered with data on that stretch, and answered badly if the windows overlap or the sample stops before the episodes that matter.

The structural point underneath it is more durable and cuts the other way: the seller of protection collects a small, frequent, capped amount and takes on an occasional large, uncapped one. That payoff shape survives whatever the average says. A strategy can be right in most windows and still be decided by the ones it was wrong in — which is the same asymmetry that makes a fall of 50% need a rise of 100% to undo.

So the honest reading of a high volatility gauge is narrow: protection is being bid, and the bid is either paying for something scheduled, something already happening, or something the market has decided to fear. Working out which of the three is analysis. The number alone does not distinguish them.

Four ways these gauges get misread

1. Reading a level as a direction. A high number says protection is expensive. Both tails contribute to it, and no arrangement of option prices contains a view about which one arrives.
2. Comparing the two levels head to head. Same units, different baselines. Each gauge is informative against its own range; the raw difference between them is mostly a fact about the two underlyings.
3. Missing the calendar inside the window. A domestic reading well above its recent range, three weeks before a Budget or a counting day, is a repriced event and not necessarily a dislocation.
4. Turning a divergence into a forecast. "The US gauge has moved and ours has not" is an observation. Everything after it — whether it transmits, when, through which channel — is a mechanism you have to be able to state, and even then the timing is not something the gauges know.

There is a fifth that costs long-horizon holders the most: watching either gauge at all when the decision in front of them has a ten-year horizon. A volatility index is a 30-day price. It has nothing to say about a monthly investment plan, and checking it daily converts a market statistic into a source of anxiety — the trap described in what happens when you stop a SIP in a drawdown.

Where this sits in the app

The volatility group on the macro page in FNOTrader's Options Analytics app has exactly two tiles in it, and they are these two. The MOVE index applies the same idea to US Treasuries but is grouped with rates and credit rather than here, on the reasoning that it is read alongside the yields and spreads it moves with rather than against an equity gauge.

Each tile carries its one-day change, a plain-English note on the mechanism it transmits through, and a history chart over one month to five years, so a reading can be placed against its own range instead of against a raw comparison with the other gauge. The composite names its largest contributors in words, and the regime label sits next to the score rather than being derived from it, so a Stress state alongside an unremarkable number is visible as exactly what it is — an override that bypassed the average.

One gap worth knowing about, because this article turns on the pair. The correlation panel runs Nifty against a fixed list of ten drivers over 30, 60 or 90 sessions, and the US VIX is on that list while India VIX is not. There is a reason it would be the least informative row on the panel — India VIX is computed from Nifty's own option book, so correlating the two would largely measure that fact rather than a relationship between two markets. It is still an absence rather than an answer, and the honest thing is to name it.

The weights, the signs and the override thresholds are the ones stated in this article. They are published rather than buried so that a reader who disagrees can see precisely what they are disagreeing with.

Common questions

Continue reading

More in Macro & Markets · App: Options Analytics · Definitions: glossary · Free tools: calculators · All: every article