Screens · Volatility
Most Volatile Stocks
A volatility screen ranks a universe by how far its members move. Two definitions are in common use, they disagree on exactly the instruments worth looking at, and neither of them says anything about direction.
The two definitions
Almost every published most-volatile list is built one of two ways, and the difference is not a refinement. The two rankings can be close to reversed on the same universe.
| Measure | Formula | Counts |
|---|---|---|
| Average true range, per cent | mean(true range) ÷ price × 100 | Distance travelled in a typical session, gaps included. |
| Return dispersion | standard deviation of daily returns | How much the daily results scatter, close to close. |
| High-low range, per cent | (high − low) ÷ close × 100 | One session's span, ignoring where it opened relative to yesterday. |
The true range is the larger of today’s high-low span and the distance from yesterday’s close to today’s extremes, which is what makes it count gaps. Return dispersion sees only closing prices, so a session that opened far away and travelled nowhere registers as a large return with no range at all.
Why the list fills with the same kind of name
Whichever measure is used, the top of the list has a recognisable character: small companies, low share prices, thin order books, and frequently something recent, a listing, a restructuring, a takeover rumour. Three mechanisms produce it, and none of them is a defect in the screen.
Percentages favour a small base. Both measures are relative to price, and a twenty-cent move on a two-dollar stock is ten per cent. The same twenty cents on a two-hundred-dollar stock is a tenth of a per cent. Nothing about the first company is more turbulent than the second; the arithmetic simply rewards the smaller denominator.
Thin books move further per order. Where few shares are resting at each price, an ordinary-sized order walks the book and the print moves several per cent. That is liquidity, not information, and it is why a liquidity floor changes the output of this screen more than any refinement of the volatility measure.
Events dominate short windows. Over twenty sessions, one earnings surprise or one takeover report can carry the entire reading. A name near the top of a short-window list is frequently a name that had one dramatic day and has been quiet since, which a 250-session window would rank nowhere near the top.
Building it yourself
Everything here runs on free end-of-day data (daily high, low, close and previous close), and the arithmetic is a few lines over a table. The work that decides whether the output is worth reading is the housekeeping around it.
- Adjusted prices only. An unadjusted split leaves a step in the series that every volatility measure reads as an enormous move. Run the screen on unadjusted data and it will faithfully return a list of recent corporate actions.
- Set a liquidity floor first. Filter on median dollar volume, share volume multiplied by price, before ranking anything, so the result is not a list of instruments where three trades set the price.
- Exclude what is not an operating company. Exchange lists carry closed-end funds, preferred shares, exchange-traded products, warrants and multiple share classes. They pass every price filter and are not what anyone screening for stocks intends to find.
- State the window and the measure beside the list. A ranking whose definition is not printed cannot be compared with anything, including with itself a month later.
Where it misleads
| Situation | What goes wrong |
|---|---|
| Definition unstated | Average range and return dispersion can rank the same universe in nearly opposite orders. A list without its definition is not comparable with any other. |
| No liquidity floor | The top of the list describes a handful of trades in a thin book rather than a market. |
| Unadjusted prices | Splits and large dividends read as enormous moves; the screen returns a list of corporate actions. |
| Volatility read as direction | The measures are symmetric. A steady, steep advance can rank as high as a collapse. |
| Short window after one event | A single earnings day can carry a 20-session reading for a month after the instrument went quiet. |
| Ranking across price levels | Percentage measures favour low-priced issues structurally, so the list is partly a list of cheap stocks. |
What volume adds
A volatility screen tells you an instrument moved and nothing about what it took to move it. The two cases it cannot separate are the ones that matter: a five per cent range on the heaviest volume of the quarter, where size was genuinely traded against, and an identical five per cent range on the thinnest session of the month, where a modest order found an empty book.
The first describes a market changing its mind. The second describes an absence of participants. Reading the volume beside the range (or ranking on both, which costs one extra column) separates them, and it is the difference between a candidate worth work and a candidate that will unwind as soon as anyone real arrives.
Frequently asked questions
What does "most volatile" actually measure?
One of two things, and the screen rarely says which. The first is average true range as a percentage of price: how far the instrument travels in a typical session, gaps included. The second is the standard deviation of daily returns: how much those returns scatter around their mean. They agree on instruments that are volatile in every sense and disagree sharply on the interesting cases, so a list published without its definition cannot be compared with any other list.
Why do small, low-priced stocks dominate the list?
Because both measures are percentages, and a percentage move is easier to produce on a smaller base. A $2 stock moving 20 cents has moved 10 per cent; a $200 stock needs a $20 move for the same reading. Add thinner order books, where a modest order moves the price further, and the top of any volatility screen is structurally a list of small, cheap and thinly traded issues. That is not a fault in the screen — it is what volatility means — but it does mean the list is rarely the list people expected.
Should I use a lookback of 20 days or 250?
Whichever matches the horizon of the decision, and state it. A 20-day window describes the last month and reacts quickly to a change in character; an earnings shock will dominate it. A 250-day window describes the year and is far steadier, but it will still be reporting last spring’s turbulence in an instrument that has been quiet since. Ranking the same universe on both and looking at the difference is more informative than either alone: a name high on the short window and low on the long one has just changed.
Does high volatility mean high risk?
It means large movement, which is not the same claim. Volatility is symmetric (it counts moves in both directions equally), so an instrument that has risen steeply and steadily can rank high on it. Whether that constitutes risk depends entirely on the position: for a short-dated option seller it is the central risk, for a long-term holder with no leverage it is mostly noise, and for someone sizing a position it is exactly the input needed to size it correctly.
Do I need intraday data to build this?
No. Daily high, low, close and previous close are enough for both measures, and both are a few lines of arithmetic over a table of daily bars. Intraday data would let you measure how the range was distributed through the session, which is a genuinely different question and the one the same-day screen answers. For ranking a universe over weeks, daily bars are the right input and free.
Why exclude illiquid issues before ranking?
Because without a liquidity floor the top of the list is occupied by instruments where a handful of trades set the price, and the reading describes those trades rather than a market. A minimum dollar volume (not share volume, which is dominated by price level) removes them. This single filter changes the character of the output more than any adjustment to the volatility measure itself.
How does volatility relate to volume?
They rise together often enough to be confused and they measure different things. Volume is how much traded; volatility is how far the price moved while it traded. The interesting cases are where they disagree: a wide range on light volume says a thin book gave way rather than that anyone large was involved, and a heavy session with almost no range says size was absorbed. That comparison is the whole method this site is built around, and no volatility screen can make it for you.
Are these lists worth screening at all?
As a starting point for work, yes; as an answer, no. A volatility screen shortens a universe to the names where something is moving, which is useful when you have a reason to want movement, an options strategy, a short-horizon system, a position-sizing input. What it cannot say is why any of them is moving, and the four common reasons (an earnings surprise, a sector rotation, an index change, and a business in trouble) call for completely different responses.
Is there a live version of this screen?
Not on this page yet, and the reason is data rather than difficulty. Running it honestly requires a daily market feed, and publishing a stale list of names would be worse than publishing none: a table dated last year reads as current to anyone arriving from a search result. What is here is the definition, the arithmetic and the pitfalls, all of which stay true, with enough detail to run it yourself on free end-of-day data.