Size by Volatility, Not by Conviction
Key takeaway
- Equal dollar amounts are not equal risk. A stock that moves 4% a day carries four times the risk of one that moves 1%, for the same money.
- Measure how far an instrument normally travels — its average true range — and let that number set the share count.
- The result is that every position can lose about the same amount, which is what makes a run of losses survivable and a track record readable.
Learning pathHow much to bet, and every way people get it wrongStep 1 of 6
Based on Systematic Trading — Robert Carver, 2015
What average true range is telling you
ATR is a measure of normal movement. It is not a prediction, and it does not say which way.
That is exactly the number a stop needs. A stop placed closer than the instrument's normal daily wander will be hit by ordinary noise, and being stopped out by noise is not risk management — it is paying the spread repeatedly for no information.
The arithmetic, which fits on one line
Shares = (what you'll risk) ÷ (stop distance). Everything else is deciding those two numbers honestly.
| Quiet stock | Volatile stock | |
|---|---|---|
| Account risk allowed | $400 | $400 |
| ATR (normal daily range) | $1.00 | $4.00 |
| Stop distance (2 × ATR) | $2.00 | $8.00 |
| Shares to buy | 200 | 50 |
| Cash committed at $50 | $10,000 | $2,500 |
Read the last two rows together. The cash committed is wildly different and that is correct — the risk is identical, which is the quantity you actually budgeted. A position sized this way looks small precisely when the instrument is dangerous.
What this changes about the equity curve
Volatility scaling rarely transforms where you end up. It transforms the path, and the path is what determines whether you are still running the method a year later.
There is a second, quieter benefit. When every position risks the same amount, your record becomes readable: a winning month means the method worked, not that one oversized holding happened to run. Without that, you cannot tell your expectancy from your luck.
Where it stops helping
- Volatility changes. The ATR you sized on is last month's; a position sized in calm conditions is oversized when things get loud. Re-check it rather than setting it once.
- It does not know about correlation. Five separately-sized energy stocks are still one bet on oil. Sizing each correctly does nothing about that.
- Gaps ignore your stop. A stop two ATRs away is not a guarantee of a two-ATR loss when the market reopens somewhere else entirely.
None of these is a reason to size by feel instead. They are reasons to treat the number as a ceiling you keep checking, which is the same conclusion the Kelly criterion reaches from a completely different direction.
Try this week
- Look up the average true range of the last stock you bought, then compare it to the stop distance you used.
- Recompute that position's share count as (risk budget ÷ 2 × ATR). Note the difference from what you actually bought.
- Do the same for your most volatile holding. Ask whether the cash figure now looks uncomfortably small — and why that discomfort is the point.
- Group your open positions by what would move them together. Size the group, not just the names.
Common questions
What is ATR in trading?
Average true range is the average distance between a period's high and low, usually over the last fourteen periods, adjusted for gaps. It measures how far an instrument normally moves, and says nothing about direction.
How do you calculate position size using ATR?
Decide the cash you are willing to lose on the trade, set a stop a fixed multiple of ATR away from entry, then divide the first number by that stop distance. The result is the number of shares.
What multiple of ATR should I use for a stop?
Two is a common starting point because it sits clearly outside normal daily movement without being so wide that the position must be tiny. The multiple matters less than using the same one consistently so results are comparable.
Is volatility position sizing better than a fixed percentage?
For risk consistency, yes: a fixed percentage of the account per position leaves the actual risk varying with each instrument's volatility. Sizing on ATR holds the loss roughly constant, which is usually what the percentage was meant to do.
Does this work for long-term investing as well as trading?
Partly. The insight that equal dollar amounts are unequal risk applies to any portfolio. The ATR-and-stop machinery is built for positions with defined exits, so a buy-and-hold investor gets more from the correlation point than from the share-count formula.