What ATR is and how it is calculated
J. Welles Wilder created the Average True Range to measure volatility alone, independent of direction. The calculation starts with each candle's "true range", the largest of three distances:
True range = the greatest of: (high − low), |high − previous close|, |low − previous close|
ATR is then the average of that range over the chosen number of periods (14 by default). Including the previous close matters: using only high minus low would ignore price gaps entirely. An instrument that gaps sharply then drifts quietly looks calm on a high–low measure, while true range captures the whole jump.
| Day | High | Low | Prev close | True range |
|---|---|---|---|---|
| 1 | 2,410 | 2,392 | 2,400 | 18 (high−low) |
| 2 | 2,455 | 2,440 | 2,405 | 50 (high−prev close) |
| 3 | 2,448 | 2,430 | 2,452 | 22 (prev close−low) |
Note day two: the candle itself looks narrow ($15 high to low) but the true range is 50 because the market gapped. That is precisely what makes ATR more accurate than measuring the candle alone.
What ATR does not measure — and its most common misreading
The most common ATR error is reading it as a directional signal. ATR knows nothing about direction. A rise means movement widened; it says nothing about whether that movement is up or down — a violently falling market and a violently rising one produce equally high ATR. Buying because ATR rose bases a decision on information that contains no direction at all.
| ATR reading | What it actually means | What it does not mean |
|---|---|---|
| High ATR | Daily movement is wide | Not a buy signal nor trend strength |
| Low ATR | The market is quiet and range-bound | Not a sell signal nor weakness |
| Rising ATR | Volatility is expanding | Does not say in which direction |
| Falling ATR | Volatility is contracting | Does not imply a reversal is near |
The second overlooked limit: ATR is denominated in the instrument's price units, so comparing it across instruments is meaningless. Gold's ATR might be $20 while EUR/USD's is about 0.0060 — the two numbers are not comparable. For a fair comparison, convert ATR to a percentage of price:
ATR% = (ATR ÷ current price) × 100
Gold at 2,400 with ATR 24 gives 1.0%; a pair at 1.1000 with ATR 0.0066 gives 0.6% — only now do you know which is relatively more volatile. Third limit: ATR is inherently lagging, being an average of what already happened, so it records a volatility jump rather than predicting one.
Application one: a stop that breathes with the market
A fixed stop (say 30 pips) treats a quiet market and a violent one identically — needlessly wide in the first, far too tight in the second, where ordinary noise takes it out. The fix is to derive stop distance from ATR:
Stop-loss = entry ∓ (multiple × ATR)
| Multiple | Stop distance (ATR = $20) | Suits | Trade-off |
|---|---|---|---|
| 1.0× | $20 | Short intraday scalping | Frequently hit by ordinary noise |
| 1.5× | $30 | Balanced day trading | A common middle ground |
| 2.0× | $40 | Swing trading | Breathing room at a smaller size |
| 3.0× | $60 | Longer-term positions | A large loss when wrong |
The mental rule: the tighter the stop the more often noise removes you; the wider it is the smaller your permitted size. There is no "correct" multiple — only one that suits your horizon, and it should stay fixed so you can evaluate results later instead of changing it after every loss.
Application two: from volatility to position size
Most explanations stop here, yet this is the link that makes ATR genuinely valuable. Having derived stop distance from ATR, the real question remains: how many lots? And the answer comes not from ATR but from your risk percentage:
Position size = (capital × risk %) ÷ (multiple × ATR × unit value)
A full gold example: a $5,000 account, 1% risk ($50), daily ATR of $20, and a 1.5 multiple giving a $30-per-ounce stop. A standard gold contract is 100 ounces, so a $1 move equals $100 on a full contract and $1 on a 0.01 contract.
| Step | Calculation | Result |
|---|---|---|
| Permitted risk | 5,000 × 1% | $50 |
| Stop distance | 1.5 × 20 | $30 per ounce |
| Permitted ounces | 50 ÷ 30 | 1.67 ounces |
| Contract size | 1.67 ÷ 100 | about 0.017 lots |
The important practical consequence: when volatility rises the stop widens automatically and your size shrinks; when the market calms the stop tightens and your size grows — while your dollar risk stays fixed at $50 in both cases. That is the fundamental difference between a trader who sizes to the market and one who always opens "0.10 lots" regardless of conditions. Run the numbers with the position size calculator, and review the principle in gold position sizing.
When not to use ATR: mistakes and limits
- Do not use it for direction or timing. ATR is a measurement tool, not a signal tool; entries need an entirely separate reason.
- Do not compare ATR across instruments directly — use ATR% or the comparison is meaningless.
- Beware ATR immediately after news: one violent candle lifts the average for several periods and makes your stop wider than the calm market that followed warrants.
- It works poorly on very small timeframes, where most of the range is spread and noise rather than real movement.
- Changing the period changes the meaning: ATR(5) reacts fast and jumps around; ATR(50) is smoother but slow to register changed conditions. The default 14 is a compromise, not a sacred rule.
- It is insufficient alone in gap-prone markets such as the post-weekend open, where price can jump past your stop regardless of the calculation.
Finally, ATR does not protect against analytical error: it controls the size of a mistake, not its likelihood. Risk management and a fixed risk percentage remain the foundation; ATR is the tool that executes them precisely.