This article explains a volatility indicator. It is educational and general, not personalised investment advice, and not a recommendation to buy, sell, or size any security. Where it references how Shishin uses the measure, that describes a research process, not a result you should expect to reproduce.
Two stocks both fall four percent and the same headline calls both “a sharp drop.” For one it was an ordinary Tuesday; for the other it was the worst session in a month. The number that tells them apart, that says how much a given name normally moves, is its Average True Range. It is one of the most useful and least glamorous figures in systematic trading, and almost everything sensible about where to place a stop or how large a position to take runs through it.
What is ATR?
Average True Range (ATR) is a volatility indicator that measures how much a security typically moves over a single period, the average, across a lookback window (commonly fourteen days), of the “true range,” where the true range for a day is the greatest of three distances: the day’s high minus its low, the high minus the prior day’s close, and the prior close minus the day’s low. Because it takes the largest of those three, ATR captures overnight gaps that a simple high-minus-low would miss, and so reflects the real magnitude of a name’s movement rather than only its intraday swing.
Read that definition slowly, because the whole value of the measure is in the word true. A plain daily range, high minus low, only sees what happened between the open and the close. It is blind to the gap: if a stock closes at 100 and opens the next morning at 92 on overnight news, then trades in a tight band from 92 to 93, its high-minus-low for that day is a sleepy one dollar. Its true range is eight, because the prior close (100) minus the day’s low (92) is the biggest of the three distances. The stock plainly moved eight dollars; the true range is the version of “range” that admits it.
True range vs. a plain daily range
That gap-handling is the entire reason “true” range exists, and it matters most for exactly the names where it is easiest to get hurt. Stocks do a large share of their real moving while the market is closed, earnings land after the bell, guidance changes overnight, a sector re-rates before the open. A range measure that ignores the gap will systematicallyunder-state how much a gappy name moves, and will do it worst on precisely the names whose moves you most need to respect.
So ATR is not just “average daily range with extra steps.” It is a deliberately conservative measure of movement: by always taking the greatest of the three distances, it never lets a quiet intraday session disguise a violent overnight one. Averaged over a couple of weeks, it gives you a single, stable number for how far the name tends to travel in a day, gaps and all, the kind of input you can actually build rules on top of, rather than a figure that lies to you every time the most important price action happens off-hours.
ATR measures magnitude, not direction
Here is the limit to internalise before you do anything with the number: ATR is a measure of magnitude, not direction. It tells you how much a stock typically moves. It says nothing whatsoever about which way. A name with a high ATR is a name that covers a lot of ground in a day, up days and down days both. A rising stock and a collapsing stock of identical volatility have the same ATR. The measure is direction-agnostic by construction, because every one of the three distances it averages is taken as an absolute size, never a signed move.
This is why ATR is never, on its own, an entry signal. It cannot tell you a stock is going up; it can only tell you how violently it is moving while it does whatever it is going to do. The direction of the trade is the job of the signal logic, trend, thrust, distance from key averages, the things that have a sign. ATR is the orthogonal question: granted you want to be in this name, how much room does it need? Confusing the two, treating high volatility as if it were a bullish tell, is one of the more expensive beginner errors, and it comes from forgetting that ATR has no opinion about up versus down.
So, what is ATR used for?
If ATR only tells you magnitude, the natural question is what that is good for. The answer is: three closely related jobs, all of which amount to scaling your behaviour to the name’s own volatility instead of to an arbitrary constant.
- Setting volatility-aware stop distances. A flat percentage stop, seven percent below entry for everything , is too tight for a high-volatility name (it gets noise-hit before any thesis plays out) and too loose for a quiet one (the loss compounds before the stop is reached). Expressing the stop as a multiple of ATR fixes both: a wider stop in dollar terms for a volatile name, a tighter one for a calm name, each set the same “normal amount of noise” away from entry. That is the whole case made in volatility-aware stops.
- Normalising position size. Once the stop distance is known in dollars, ATR lets you size the position so that hitting the stop costs roughly the same amount on every trade, regardless of how jumpy the name is. A volatile name gets a wider stop and therefore a smaller share count; a quiet name gets a tighter stop and a larger one. The dollar at risk per position is held roughly constant, a discipline that pairs with sizing by conviction, where the rank of the idea decides how much to commit and the volatility decides how that commitment translates into shares.
- Gauging expected daily movement. Divide ATR by the price and you get a rough Average Daily Range as a percent , a quick read on what a “normal” day looks like for this name. A two-percent-ADR stock and a six-percent-ADR stock are different animals; the same dollar move means panic in one and a shrug in the other. Knowing which kind of name you are holding is half of reading any setup honestly, and it feeds straight into the volatility inputs of a composite score.
How Shishin uses ATR
Shishin treats ATR as plumbing rather than as a headline. Every entry’s protective stop is derived from the name’s recent ATR rather than a fixed percentage. The consequence is the one the textbook promises but most retail systems never actually wire up: a high-volatility name is given a wider stop while carrying the same dollar risk as a quiet one, because the position is sized down to compensate. Equal dollar at risk per name is the design target; differing stop widths and differing share counts are simply the arithmetic that gets you there.
Two details matter, because they are where a real system differs from a diagram. First, the ATR-derived stop is set once, at entry, and then left alone, it is not ratcheted tighter bar by bar as the trade moves. The volatility read defines the room the name needs at the moment of commitment; constantly re-tightening it tends to convert ordinary oscillation into premature exits. Second, the volatility-derived distance sits under an upper cap: above a certain level of volatility the implied stop would be so wide that the worst-case loss is no longer one the system is willing to underwrite, and the name is simply treated as un-tradeable for that setup rather than entered with a reckless stop. The names this bites hardest on differ by guardian, a small-cap-momentum engine like Suzaku meets far more high-ATR candidates than a defensive-sector engine like Byakko, but the principle is shared across the stack. What we publish is that principle; the specific multiples, caps, and lookbacks are part of the engine’s edge and stay unpublished.
What ATR does not tell you
ATR is honest about magnitude and silent about everything else, and it is worth being equally honest about its limits. It is backward-looking: it is an average of what already happened, so it describes the recent past and quietly assumes the near future rhymes with it. It says nothing about direction, as covered above, so it can never be the reason to be long or short. And a low ATR is a statement about ordinary days only, a quiet, low-ATR name can still gap violently on news, and when it does, the realised move can dwarf anything its placid recent range would have suggested. ATR sizes for the weather, not the earthquake.
None of that is a flaw in the measure; it is the measure doing exactly its one job and no more. The mistake is asking ATR to be a forecast or a direction signal. Asked the question it was built for , how much does this name typically move?, it is one of the most reliable inputs in the entire toolkit, which is why so much of disciplined risk management is quietly built on top of it.
So, what is ATR used for, in one line?
Average True Range answers a single question, how much does this security normally move, gaps included, and answers it well. That one number is what lets a system place a stop at the right distance for the name rather than an arbitrary one, size each position so a loss costs about the same everywhere, and tell a sleepy stock from a frantic one. It will never tell you which way a price is going. It was never supposed to. Used for what it is , a clean, evergreen measure of magnitude, it quietly does more honest work than most of the indicators that get top billing.
Sources & further reading
- Wilder, J. W. (1978). New Concepts in Technical Trading Systems. Trend Research.