Volatility-based position sizing sets share count inversely to an instrument's volatility so that every position contributes roughly the same dollar risk — a risk-parity idea applied at the single-trade level. Instead of buying a fixed number of shares or a fixed dollar amount, you first measure how much the stock moves (almost always via the Average True Range, ATR), express your stop in ATR units, and then solve for the share count that makes a stop-out cost a fixed slice of equity. A high-ATR name gets fewer shares; a quiet name gets more. The core tension: it equalizes risk across a watchlist of different-volatility stocks, but it does not size for edge — a high-conviction setup and a coin-flip setup receive the same risk unless you deliberately overlay conviction on top.

How it's calculated

The canonical single-name formula:

Shares = (Equity × Risk%) / (ATR-multiple × ATR)

• Equity — account value being risked.
• Risk% — fraction of equity you accept losing if the stop hits (commonly cited as 0.5%–2% per trade; 1% is a frequent default — see Fixed-Fractional Sizing).
• ATR — Average True Range, typically the 14-period ATR for swing work (Wilder's original default), in dollars per share.
• ATR-multiple — how many ATRs of adverse move your stop allows (commonly 1.5×–3×). The product ATR-multiple × ATR is the stop distance in dollars per share.

The denominator is just the per-share dollar loss at the stop, so the formula is identical to fixed-fractional sizing with an ATR-defined stop: Shares = dollar risk / per-share stop distance. The "volatility-based" character comes entirely from letting ATR set the stop distance — wider stops on volatile names mechanically shrink the size.

Turtle "N" unit sizing (the canonical example). The Turtles (Dennis/Eckhardt) made this explicit. They defined N as the 20-day exponential average of True Range — their volatility measure. Dollar Volatility = N × dollars-per-point (the dollar move of one contract per 1 N of price move). A Unit was sized so a 1 N move equals 1% of account equity:

Unit = (1% × Account) / Dollar Volatility = (1% × Account) / (N × $/point)

Stops were placed at 2N from entry, capping risk at ~2% of equity per unit. Exposure was capped at 4 units per market, 6 units across closely correlated markets, and roughly 10–12 units total across the portfolio (limits cited from the Original Turtle Trading Rules). This is the same Equity × Risk% / (multiple × ATR) structure, expressed in units and points rather than shares.

How it's used in practice

• Normalizing a watchlist. The dominant use: line up several candidates of very different volatility and size each so a stop-out costs the same dollars. Without this, a fixed-share or fixed-dollar approach silently over-risks the volatile names.
• Adaptive sizing as volatility shifts. Because ATR is recomputed continuously, the same stock gets a smaller position when it's volatile and a larger one when it's calm — the system leans into quiet markets and out of turbulent ones automatically.
• Portfolio-level extensions. The same logic scales up. Inverse-volatility weighting sets each holding's weight proportional to 1/volatility (an asset at 10% vol gets twice the weight of one at 20%). Equal Risk Contribution (ERC) / risk parity generalizes this to account for correlations so each position contributes equally to total portfolio variance — not just equal standalone vol. Volatility targeting scales gross exposure (adding leverage or cash) to hold the whole portfolio at a fixed annualized vol target. These are the institutional cousins of the single-trade ATR sizing rule.

Standing & evidence

Volatility-based sizing is well established and uncontested as a risk-equalization technique — it is standard in CTA/managed-futures and risk-parity mandates, and ATR sizing is a staple of trend-following and swing-trading curricula (Van Tharp catalogs it as the "percent volatility model," one of his core position-sizing methods). What is not established is that it improves returns: it is a risk-allocation rule, not an alpha source. Risk-parity's empirical record is mixed and regime-dependent — it can underperform when low-vol assets are richly valued and is vulnerable to correlation breakdowns, where assets that looked independent move together in a crisis and "equal risk" turns out concentrated. Volatility targeting has been documented (e.g., ECB financial-stability work) to be pro-cyclical: many funds cutting exposure on the same vol spike can amplify a sell-off. Treat the method as a disciplined way to control risk, not a way to make money.

Strengths & limitations

Strengths. Equalizes dollar risk across heterogeneous instruments; prevents the classic error of oversizing a volatile name; ties size mechanically to a defensible, market-defined stop; removes a layer of discretion (and emotion) from the size decision.

Limitations / failure modes.

• Backward-looking. ATR and realized volatility describe the recent past. A gap or vol spike can make yesterday's "correct" size instantly too large — the position you sized for a calm tape is now oversized for a violent one. The stop you set in ATR terms can also be jumped entirely (overnight gaps), so actual loss can exceed Risk%.
• Sizes for parity, not edge. It deliberately ignores expected value. A great setup and a marginal one get the same risk unless you overlay conviction (e.g., scale Risk% by a conviction tier) — risk parity alone leaves edge on the table.
• Whipsaw from over-tuning. Recomputing size on every ATR tick, or using too short an ATR lookback, makes position size jittery and can churn entries/exits. Most practitioners fix size at entry rather than resizing continuously.
• #1 misuse: treating it as a profit engine or a substitute for an edge. It is a defense — it caps how wrong any single trade can go; it does not tell you whether to take the trade.

Worked example

Equity $50,000, Risk% = 1% ($500 risk). Stock A: ATR(14) = $2.00, stop = 2× ATR. Stop distance = 2 × $2.00 = $4.00/share → Shares = $500 / $4.00 = 125 shares (~$ position depends on price). Stock B: ATR(14) = $0.50, same 2× stop. Stop distance = $1.00/share → Shares = $500 / $1.00 = 500 shares. Both positions lose ~$500 if stopped — equal dollar risk despite B getting 4× the share count, because B is one-quarter as volatile.

System relevance

This node is the volatility-normalized member of the Position Sizing family. It pairs naturally with [[ATR Stops]] — that node defines the stop distance this formula consumes — and is a direct variant of [[Fixed-Fractional Sizing]] (same risk / stop-distance math, with ATR supplying the stop). For Augustus, the practical wiring: use volatility-based sizing to put each watchlist candidate on equal risk footing, but apply a conviction overlay (scaling Risk%) before committing, since the raw method is edge-blind. Always flag to the agent that ATR is backward-looking — a recent vol spike means the current ATR may understate forward risk, and size should be cut, not trusted, into expanding volatility.

Sources

• Original Turtle Trading Rules (Dennis/Eckhardt system) — N as 20-day exponential average of True Range; Dollar Volatility; 1% Unit; 2N stop; unit limits (4/market, 6 correlated, ~12 total).
• QuantifiedStrategies — "Position Sizing in a Turtle Trading System" and "Volatility-Based Position Sizing."
• Van Tharp, Trade Your Way to Financial Freedom — percent-volatility position-sizing model (via Vantharp Institute / Zerodha Varsity summaries).
• Investopedia / StockCharts ChartSchool — ATR (Wilder, 14-period default) and stop-distance sizing.
• Quantilia / QuantInsti / Wikipedia "Risk parity" — inverse-volatility weighting and Equal Risk Contribution definitions.
• ECB Financial Stability Review (2020), "Volatility-targeting strategies and the market sell-off" — pro-cyclicality of vol-targeting.