← Back to blog

Volatility-Scaled Sizing: How to Size Trades by Risk

November 18, 2026
Volatility-Scaled Sizing: How to Size Trades by Risk

Volatility-scaled sizing means you make a position smaller when an asset swings harder and larger when it is calm, so every trade puts about the same amount of your account at risk. A $10,000 position in a coin that moves 5% a day is not the same bet as $10,000 in a stock that moves 1%. Scaling by volatility fixes that. It works the same way for crypto, forex pairs and stocks, and it is the core of how disciplined traders keep one wild market from deciding their month.

If you already use a fixed percentage of your account per trade, this is the next step. You keep the percentage and let the market's current volatility set the stop distance and, with it, the size.

Disciplineaiapp
disciplineaiapp.com
Size Trades With Risk in View
Discipline AI brings market-condition analysis, trade planning, and performance analytics together to support more informed, risk-aware trading decisions.
Visit

TL;DR

  • Volatility-scaled sizing sets position size so a normal move in the asset costs you roughly the same dollar amount, whatever you trade.
  • The practical formula: size = dollar risk ÷ (volatility measure × stop distance in multiples of it). ATR is the common volatility measure.
  • Risk budgeting divides your total risk across positions. Risk parity is the special case where each position gets an equal share.
  • Volatility is measured from the past, so it lags regime changes. Cap leverage, set a maximum size and review results on a real sample.
  • Bayesian sizing shrinks size when your evidence is thin and lets it grow as trade results accumulate.

Table of Contents

What volatility-scaled sizing actually does

Volatility is how far an asset's price typically moves over a period, usually the standard deviation of returns. Two ideas sit under the name, and the terms are often used interchangeably.

Volatility normalization puts returns on a common scale. The Trend Following Systems documentation defines a normalized return as the return divided by the previous period's volatility estimate. It uses yesterday's estimate so that today's move does not scale itself (Trend Following Systems).

Volatility-scaled (or volatility-targeted) sizing applies the same division to your exposure. The same source gives the target-volatility weight as the target volatility divided by the asset's current volatility (annualized), multiplied by your trading signal. It also notes that the target is only approximate, because the volatility figure is an estimate.

In plain terms, with a 15% annualized volatility target:

  • An asset with 60% annualized volatility gets 15 ÷ 60 = 25% of your equity.
  • An asset with 30% annualized volatility gets 15 ÷ 30 = 50% of your equity.

Both positions are meant to contribute the same expected fluctuation to your account. The calmer asset simply needs more capital to do it. Keep this in mind for the rest of the article: you are not predicting direction, you are controlling how much each trade can hurt.

Two ways to calculate it

Method 1: Risk per trade and an ATR stop

Most active traders do not run annualized target math on every trade. They size from the stop. The average true range (ATR) measures the typical daily price range of an asset over a lookback, commonly 14 periods. The routine:

  1. Decide the dollar risk: account equity × risk percentage. A common range is 0.5% to 1%, covered in our guide to position sizing methods for active traders.
  2. Measure the asset's ATR.
  3. Place the stop a set multiple of ATR away, for example 1.5 × ATR.
  4. Size = dollar risk ÷ stop distance.

The stop distance is what carries the volatility. A wild asset gets a wide stop in dollar terms, so the same dollar risk buys fewer units.

A worked example

These figures are hypothetical, to show the arithmetic. Say your account is $25,000 and you risk 1% per trade, or $250.

Bitcoin (hypothetical)Large-cap stock (hypothetical)
Price$80,000$190
14-day ATR$2,400 (3.0%)$2.85 (1.5%)
Stop at 1.5 × ATR$3,600$4.28
Size at $250 risk0.069 BTC58 shares
Position valueabout $5,550about $11,020

The stock position is about twice the value of the bitcoin position, yet both lose about $250 if the stop is hit. A flat "put $10,000 in each" rule would have risked far more on the coin.

Forex works the same way, with pips instead of dollars. You convert the stop distance in pips into account currency using the pip value of the pair. Our forex position sizing formula walks through that conversion.

Method 2: Target volatility on the whole portfolio

The second method is closer to what funds do. You pick a target for your overall volatility and set each position's weight as target ÷ forecast volatility. It suits swing traders and investors who hold several positions for weeks. Our article on volatility targeting covers how to set the target and rebalance.

Choosing the volatility window

The lookback you pick changes the answer. A short window, such as 10 days, reacts quickly to a sudden spike but is noisy, so your size jumps around. A long window, such as 90 days, is steadier but slow to notice that the market has changed character. Many traders land on a 20 to 60 day window for daily data, or run two windows and use the larger reading as a guard.

A few practical points that trip people up:

  • Annualizing. Stock and forex traders usually multiply daily volatility by the square root of roughly 252 trading days. Crypto trades every day, so many crypto traders use 365. Use one convention consistently so numbers compare.
  • Use returns, not prices. Volatility of percentage returns is comparable across assets. Raw price changes are not.
  • Timeframe match. If you hold for hours, daily ATR is the wrong measure. Use the ATR of the timeframe you trade.

Risk budgeting and risk parity

Once you hold more than one position, a new question appears: how much of your total risk goes to each? The CFA Institute's review of the subject gives the cleanest answer. Risk budgeting is the general practice of allocating risk across a set of assets. Risk parity is a special case that allocates risk equally (Financial Analysts Journal).

For a trader, that translates into three levels:

  1. Per-trade budget. The 0.5% to 1% you risk on any one idea.
  2. Portfolio budget. A cap on total open risk, for example 4% to 5% of equity across all trades at once.
  3. Group budget. A cap on correlated trades, such as several altcoins that all move with bitcoin, or three USD pairs that all bet on the same dollar move.

The simple version of parity weights each asset by the inverse of its volatility, so calmer assets get more capital. That is the same logic as the 15% target example above. It ignores correlation, which is the main reason it is an approximation: two assets with the same volatility that move together are one risk, not two. In crypto, correlations tend to rise when the market sells off, which is exactly when you most need the budget to hold. Treat correlated positions as a single position when you count your open risk.

Bayesian sizing: let evidence set the size

Volatility tells you how big the market's swings are. It does not tell you how much to trust your own edge. Bayesian position sizing handles that second problem.

The idea is simple. You do not know your true win rate for a setup. You only have a sample of results. A Bayesian approach treats the win rate as a range of plausible values that narrows as trades accumulate, then sizes down when that range is wide. A research paper on a Bayesian version of the Kelly criterion, a classic growth-optimal sizing formula, builds this around uncertainty in the win probability and reports, in simulation, smaller drawdowns than full Kelly sizing (SSRN). Simulations are not live results, but the logic carries over to how you size.

In practice:

  • 10 trades with 7 wins looks like a 70% setup, but the plausible range is huge. Size it small.
  • 150 trades with 100 wins is a 67% setup with a narrow range. You can size up toward your normal budget.
  • If recent trades drift away from your historical win rate, the range widens again and size steps back down.

This is also why sample size matters more than any single impressive streak. We go deeper on turning probabilities into size in our guides on confidence-weighted sizing and calibrated probabilities.

Where volatility scaling goes wrong

Volatility scaling reduces surprise. It does not remove it.

Volatility is backward-looking. A 30-day reading describes the last 30 days. When a market shifts regime overnight, such as a surprise macro release, an exchange failure or a flash crash, your size was set for conditions that no longer exist. Gaps also jump past stops, so your realized loss can exceed the dollar risk you planned.

Low volatility invites oversizing. A quiet market produces a tight stop and a large position. If you use leverage, a single volatility jump can wipe the margin cushion. Set a hard maximum position size, as a percentage of equity, that overrides the formula, and set a maximum leverage.

Liquidity is not in the formula. Thin altcoins and off-hours forex can slip well past your stop. Account for fees and slippage when you calculate risk.

Backtests flatter. A white paper on backtesting stresses that historical simulations can mislead through optimistic assumptions, ignored transaction costs and overfitting. It recommends including fees, slippage and liquidity costs and checking out-of-sample results (Daemon Investments). It also says simulations do not guarantee returns or cap your maximum loss.

A sensible way to test volatility scaling in your own trading:

  1. Define your rules in writing before you test: window, stop multiple, risk percentage, caps.
  2. Run them on one period of data, then check on a separate period you did not tune on.
  3. Compare maximum drawdown and the size of your worst day, not just total return.
  4. Practice the rules live without capital before risking any.

What this looks like day to day

The formula is the easy part. The hard part is following it when a setup looks great and the math says to take a small size, or when volatility spikes and the math says to cut a position you like. Most sizing failures are discipline failures: widening a stop, doubling up after a loss, or ignoring the budget on a "sure thing."

That is why it helps to have the numbers done for you before you click buy, and to keep a record you can review. A journal that logs the planned risk next to the actual risk shows quickly whether you really size the way you think you do. If you are newer, our list of risk controls for new traders is a good companion to this article.

Putting it to work with Discipline AI

Discipline AI is a mobile app for crypto, forex and stock traders, on iOS and Android. It includes risk management tools with position sizing and leverage calculation, so the size comes from your risk budget and stop distance rather than from a feel for the chart. Its Sentinel research system evaluates volatility, trend direction and market regimes across multiple cryptocurrencies and timeframes, and helps you see when conditions call for caution or standing aside.

Putting it to work with Discipline AI

You can rehearse the routine without risking capital. Paper trading tracks your P&L, and market replay lets you practice entries with the future hidden. Automated journaling and performance analytics then show how your actual sizing compared with your plan. We do not promise profitable trades. The aim is measurable evidence about your own decisions.

Pro is $8.99 per month, $79.99 per year or $199.99 one-off, as listed on the pricing page.

Disciplineaiapp

If you want the sizing, journaling and review habits taught step by step, The Disciplined Trader is available for a one-off $79.

Perspective

Most traders think they have a strategy problem. Look at their journals and you usually find a sizing problem. The same setup, risked at 0.5% one day and 3% the next, produces results nobody can learn from.

We built Discipline AI around evidence rather than predictions. Volatility-scaled sizing fits that approach because it asks a question you can measure: how much can this trade realistically cost me, and is that inside my budget? Sentinel studies volatility, market regimes and historical outcomes so that question has an answer grounded in data, and the app's position sizing and leverage tools turn it into a number you can act on.

Just as important is knowing when not to trade. When conditions are unfavorable, standing aside is a position too. Our stand-aside protection exists for that moment, and the journal and performance analytics show you afterward whether you followed your own rules. Sample size, risk-adjusted expectancy and reward-to-risk are what tell you whether a method deserves more capital, and we would rather you earn that conclusion slowly than assume it quickly.

Start small, size by risk, and let your own results set the pace.

Tony

FAQ

What is the difference between volatility-scaled and volatility-normalized sizing?

Normalization divides returns by a volatility estimate so different assets can be compared on one scale. Scaled sizing applies the same division to exposure, so a more volatile asset gets a smaller position. Many traders use the terms for the same practice, and both use the previous period's volatility so a trade does not size itself from its own move.

Is risk parity the same as a risk budget?

No. Risk budgeting is the general act of dividing your total risk among positions, in any proportions you choose. Risk parity is the special case where each position gets an equal share of risk. You can run a risk budget that favors your best-evidenced setups and still be doing risk budgeting, not parity.

What volatility window should I use?

Short windows react fast but are noisy, and long windows are steady but slow to notice a regime change. Many traders use 20 to 60 days on daily data, or compare two windows and take the more cautious reading. Whatever you pick, write it down and keep it fixed while you test.

Does Bayesian sizing mean I should bet more after a winning streak?

Not by itself. A short streak barely narrows the uncertainty about your true win rate, so it justifies only a small increase. Size should grow as the number of trades behind your results grows, and shrink if recent trades move away from your history.

Does volatility scaling protect me from big losses?

It keeps normal-day risk consistent, but it cannot see a regime change coming, and gaps can pass through your stop. Pair it with a maximum position size, a leverage cap and a limit on total open risk. None of these guarantees against loss.

Sources