Volatility in Equity Markets: What It Measures and What It Does Not

Volatility measures how much a price moved about its average over a stated period, usually as the standard deviation of its returns. Volatility is a measure of dispersion, and it treats a rise and an equal fall identically. The measure records what a price did. A volatility figure carries no information at all about whether a holder can lose money.

Underneath that sits one arithmetic step, and once it has been seen the rest of this guide follows. Somewhere in the calculation every daily move gets squared. A move of plus 1.77 per cent and a move of minus 1.77 per cent both become 3.1329, and from that instant onward the measure has no way of telling them apart. Everything volatility can say, and everything it cannot, comes out of that one line of arithmetic.

What is volatility actually measuring?

Think about two shopkeepers on the same street, each of whom takes home Rs 40,000/- in an average month. The first takes home almost exactly Rs 40,000/- every month, year in and year out. The second takes home Rs 12,000/- in some months and Rs 70,000/- in others, and it still averages Rs 40,000/-. Nothing about the average separates them. The two shopkeepers are separated by how widely the monthly figure scatters around that average, and that scatter is the only thing volatility is built to report.

Volatility is the dispersion of returns around their average over a stated period, and dispersion is the whole of it. Not the level of the price, not the direction of travel, not the destination. Only the width of the scatter. Look again at the two panels above. The average daily move is identical in both. The right panel is simply wider, and every difference between the two figures printed underneath them comes from that width alone.

The second thing to hold on to is that the measure is symmetrical. A big rise contributes exactly as much as a big fall of the same size. The symmetry is not an approximation or a simplifying assumption anybody made for convenience. Squaring every deviation is what produces it.

How is it computed from a price series?

Six steps, and none of them is difficult. Take the returns period by period. Average them. Subtract that average from each return to get a deviation. Square every deviation. Average the squares. Take the square root of that average. The square root is the standard deviation of the returns, and that number is what people call volatility.

Here is the ladder run on five made up daily moves, small enough to hold in mind while the six steps are followed.

The 16.40 divided by the five days gives 3.28. The square root of 3.28 is 1.81, so this five day series has a daily standard deviation of 1.81 per cent. Two things stand out. Day four and day one are almost the same size in opposite directions, minus 2.3 and plus 2.3, and both land on 5.29. And the deviations added up to nought, as they always do. Squaring is what has to happen to them before they can be averaged into anything useful.

There is a second convention hiding in step five. Dividing 16.40 by five, the number of observations, is what has been done here. Dividing instead by four, one fewer than the number of observations, gives 4.10 and a standard deviation of 2.02 per cent. Same five returns, two defensible answers, a difference of about 0.21 percentage points. Nobody looking at a published volatility figure can tell which of the two divisors produced it unless the writer says so.

Why is the annual figure multiplied rather than measured?

The number almost everyone quotes is an annual one, and almost nobody measured it. Sarvani Coatings' record gives a daily standard deviation of about 1.77 per cent. To turn that into an annual figure, the convention is to multiply by the square root of the number of trading daysA day the exchange is open for dealing. Weekends and exchange holidays are not trading days. The count in a year is nearer 250 than 365. in a year. Take 250 days. The square root of 250 is 15.8114. Multiply: 1.77 times 15.8114 gives 27.9862, or 27.99 to two decimal places and 28.0 to one.

Nothing was observed for a year. The annual figure is built from a daily measurement and a rule about how to scale it. The practice is normal and useful. A reader handed 28.0 per cent on its own will assume a year of something was watched, so the multiplication is worth showing every time.

Why is volatility not a measure of risk of loss?

Risk is the word attached to this number so routinely that the substitution has stopped feeling like one. Sarvani Coatings rose from an illustrative Rs 402/- to an illustrative Rs 486/- over the twelve months to 28 August 2026, a price return of 20.90 per cent, and with the Rs 4.00/- dividend paid in between a total returnThe return a holder actually received, counting both the change in price and the cash paid out along the way. Total return is covered separately. of 21.89 per cent. Its volatility over that period is about 28.0 per cent a year. Not one rupee was lost by anyone holding it start to finish.

Now take that same price path and reflect it. Every daily move keeps its size and reverses its sign, so the share falls from Rs 402/- to Rs 332.52/-, a price move of minus 17.28 per cent. The deviations are identical in size, so when they are squared they are identical full stop, and the reflected path has a volatility of exactly 28.0 per cent as well. The measure answers how much the price moved, never which way, and a holder who did well and a holder who did badly can report the same volatility figure to the second decimal place.

What does a high volatility figure actually establish?

Something real, and worth having. A high figure establishes that the price moved about a good deal over the period measured. Movement of that size bears directly on what a holder would have experienced along the way, on how far the quoted price could sit from where it was last checked, and on how much a position might be worth at an arbitrary moment rather than at a chosen one. When a holder has to sell on a date somebody else picked, the width of the scatter is exactly the thing worth knowing.

Look at how large the wobble is next to the drift. Sarvani Coatings' price went from Rs 402/- to Rs 486/-. Across 250 days that works out at a compounding move of about 0.0759 per cent a day. The daily standard deviation is 1.77 per cent. The typical day's wobble is about 23.3 times the average day's drift, so a single day, a single week or a single month tells almost nothing about the year. The direction is buried under the noise at short horizons and only emerges once enough days have been added together.

What does the figure depend on that is not obvious?

Three things, and none of them is normally printed beside the number. The first is the measurement period: a window that happened to contain a results day, a large corporate announcement or a broad market shock produces a different figure from a quiet stretch of the same length. The second is the sampling frequency: daily returns, weekly returns and monthly returns are three different series taken from one price history, and they do not have to agree. The third is the annualisation convention, meaning the day count someone chose.

The day count is the easiest to demonstrate because it is pure arithmetic on the same 1.77 per cent. On 240 days the annual figure is 27.42 per cent. On 250 it is 27.99. On 252 it is 28.10. On 260 it is 28.54. Identical data, four defensible conventions, a spread of 1.12 percentage points, and not one of the four answers is wrong.

The frequency is subtler and more interesting. Under the standard convention, a weekly standard deviation is implied by the daily one at 1.77 times the square root of five, or 3.96 per cent. Annualising that over 50 weeks gives 3.96 times 7.0711, back on 27.99 per cent. The two routes agree perfectly, and they agree because the scaling rule assumes each step is independent of the one before it. On an actual price series that assumption does not have to hold, the measured weekly standard deviation is not obliged to equal 3.96 per cent, and the two routes then land in different places. The agreement in the diagram below is manufactured by the convention rather than observed in the data.

How does somebody actually use this number in practice?

Start with the household version of the same idea. A salaried person with a steady monthly credit can plan a rent payment against next month's income with near certainty. A street vendor whose takings swing between Rs 400/- and Rs 2,800/- a day has the same monthly average and cannot. The vendor is not poorer, and is not more likely to fail. The vendor simply cannot commit to a payment on a fixed date without holding a cushion. The scatter, not the average, is what decides how large that cushion has to be.

Meghna Iyer, an analyst covering Sarvani Coatings, uses the figure in three narrow ways and refuses it in a fourth. She uses it to size how far the quoted price might sit from her last check, and that tells her how stale a price in a draft note is likely to be. A two per cent day on a share that typically moves 1.77 per cent is ordinary and a ten per cent day is not, so she uses it to judge how much of a single day's move is worth investigating. At 1.77 per cent a day against 0.0759 per cent of drift, a one month observation carries almost no signal about a view built on a year, so she uses it to set the horizon of a claim. And a question about whether the share is risky is a question about losing money, which this number is not about, so she refuses it entirely.

The practitioner's discipline is not that the figure is unreliable, it is that the figure answers a narrow question very well and a broad one not at all. Notice too what she never does with it: she does not combine it with the capitalisation classification, the free floatWhatever part of the share register is actually available to change hands, once promoter stakes and other locked holdings are set aside. Covered separately. or the traded value to manufacture a single ranking. Sarvani Coatings' market capitalisation is an illustrative Rs 11,664 crore with a free float of 47.6 per cent, or Rs 5,552 crore. Capitalisation, free float and volatility answer different questions, and stacking them into one score would destroy the meaning of each.

How should a volatility figure be written down?

With four items of context attached to it, every time. The period measured, so a reader knows what window produced it. The sampling frequency, so a reader knows whether these are daily, weekly or monthly returns. The annualisation convention, meaning the day count used, so a reader can reproduce the multiplication. And the as of date. A figure computed on a window ending last March is a different object from one ending today.

Written properly, the Sarvani Coatings figure reads: annualised volatility of about 28.0 per cent, from daily returns over the twelve months to 28 August 2026, annualised on 250 trading days. The written form is one number and twenty three words of context. Needing four items of context to make one number safe is itself the clearest warning about quoting it alone. Volatility 28 per cent and nothing else is a figure a reader cannot check, cannot reproduce and cannot compare with anything.

Where did these numbers come from?

Every figure attached to Sarvani Coatings is assumed rather than observed. Its daily standard deviation of about 1.77 per cent is a chosen starting point, and a share whose daily moves ran at half that size would carry half the annual figure. The arithmetic sitting on top of it is exact. The annualisation is worked step by step, and every line of it can be repeated with a calculator to give the same answer. A real calculation starts from a published price history, and the exchanges below are where it is published.