Two Stocks, Same Volatility, One Never Cost You a Rupiah: What Volatility Actually Measures
6 min read
Loading...
6 min read
"Volatile" gets used as a synonym for "risky" so often that the two feel interchangeable. They are not. Volatility is a measurement of how much a price swings, in either direction. It has no opinion on whether those swings ended up costing you money. A stock can be highly volatile and finish the year up. Another stock, with the exact same volatility number, can finish the year down badly. The number alone cannot tell you which one you are looking at.
Volatility, in the form used across finance (and in NetWort), is the standard deviation of an asset's daily returns, annualised. It measures dispersion: how far, on average, daily moves scatter from the average daily move. A sharp 5% rally and a sharp 5% crash contribute exactly the same amount to that number. Volatility cannot distinguish them, because by construction it is not built to.
Maximum drawdown is the metric that actually answers "how much did I lose": the largest peak-to-trough decline in an asset's history, in one direction only. A previous piece on this site covers it in full. Volatility and drawdown are often read as if they measure the same thing because high-volatility assets frequently also have large drawdowns, but the relationship is not automatic. It is entirely possible for two assets to carry identical volatility while one has barely any drawdown at all and the other has lost more than half its value.
Take two hypothetical stocks, A and B, each starting at an index value of 100. Over six months, Stock A rises 8%, falls 5%, rises 9%, falls 4%, rises 10%, falls 3%. Stock B moves by the exact same amounts every month, just with every sign flipped: falls 8%, rises 5%, falls 9%, rises 4%, falls 10%, rises 3%.
| Month | Stock A | Stock B |
|---|---|---|
| 1 | +8% | -8% |
| 2 | -5% | +5% |
| 3 | +9% | -9% |
| 4 | -4% | +4% |
| 5 | +10% | -10% |
| 6 | -3% | +3% |
Because Stock B's monthly returns are the exact mirror of Stock A's, both series have precisely the same standard deviation: roughly 7.2% per month, or about 24.9% annualised using the same annualisation NetWort applies to its own volatility figures (multiplying by the square root of the number of periods in a year). Any volatility screen, sorted by that single number, would rank these two stocks as identically risky.
Compounded month over month, Stock A ends the period at roughly 114.6, up about 14.6%. Stock B ends at roughly 84.7, down about 15.3%. Same volatility. Thirty percentage points apart in outcome.
Because the plain volatility figure can hide exactly this, NetWort's own asset pages do not stop at one volatility number. Each asset detail page shows a "bumpiness" card with two figures side by side: total volatility, the standard deviation described above, and downside volatility, the standard deviation of only the days with a negative return. When downside volatility comes in meaningfully lower than total volatility, specifically below 70% of it, the page flags the asset with "Most volatility is upside": a real, if unusual, situation where an asset swings hard but most of that swinging has been in its favour.
This distinction only works because it is computed the same way NetWort computes ordinary volatility underneath: an annualised standard deviation of daily returns, using sample standard deviation (dividing by n minus one rather than n) rather than population standard deviation, which the codebase treats as a fixed correctness rule precisely because the wrong version quietly understates true volatility. Downside volatility applies that identical calculation to a filtered subset of days, the negative ones only, which is also the building block behind the Sortino ratio shown alongside it.
Bitcoin is a useful real case because its volatility number is large enough to be unambiguous on its own: NYU Stern's V-Lab put its GARCH-model one-month forecast at 47.90% annualised as of 7 August 2026, several multiples of a typical large-cap stock's volatility. That number, by itself, says nothing about whether the current stretch has been a good one for holders or a bad one. Reading that requires the same second step as the hypothetical example above: checking the downside split or the drawdown, not the headline volatility figure.
IHSG's own 2026 stretch shows the other side of the same coin. Its swing from a record 9,133.87 on 19 January 2026 down to 6,409.65 by 7 August 2026 was both highly volatile by any measure and, unlike the Bitcoin figure above, a period where the volatility really was mostly to the downside. The number "volatile" fit both situations. Only the direction of the swings, not the volatility figure itself, tells you which one you were in.
A high volatility number on its own is not a verdict. It says a position moves a lot. Whether that movement has mostly worked for you or against you is a separate question, one that total volatility cannot answer and drawdown or a downside-volatility split can. Sharpe ratio, the risk-adjusted return figure NetWort also shows, is built directly on top of total volatility as its denominator, which is exactly why a single Sharpe number, like a single volatility number, benefits from being read alongside the downside picture rather than in isolation. Beta is a related but different question again: not how much a price swings in isolation, but how much of that swinging tracks a benchmark like IHSG.
The same caution that applies to reading your portfolio's return number applies here: the return your broker shows you is not automatically your real return, and the volatility number a screener shows you is not automatically a loss number either. Both require one more layer of reading before they mean what they seem to mean.