Volatility targeting means you fix the amount of risk you want to carry and let the size of your position float to match it. When the market is calm, you hold more. When it turns violent, you hold less. The rule is simple: exposure equals your target volatility divided by the volatility you measure. This guide shows the math, a worked example, the choices that matter in crypto, forex and stocks, and the places where the method breaks.
TL;DR
- Volatility targeting keeps risk roughly constant by scaling exposure up when volatility is low and down when it is high, as Man Group describes.
- Position size is target volatility divided by forecast volatility, with a leverage cap and a floor at zero.
- The evidence on returns is mixed. The risk control is the more dependable benefit.
- It reacts after the fact, so gaps, slow declines and transaction costs can hurt. Stops and drawdown limits still matter.
- Test it on paper, with costs included, before you use real capital.
Table of Contents
- What volatility targeting actually does
- The formula, step by step
- Choosing a volatility estimate
- What the research says about benefits
- What it looks like in crypto, forex and stocks
- Where volatility targeting fails
- How to implement it without overtrading
- Testing before you risk capital
- Perspective
- Put your sizing rules into practice
- FAQ
- Sources
What volatility targeting actually does
Most traders size by conviction or by a fixed dollar amount. Volatility targeting replaces that with a risk budget. Man Group's research describes the approach as targeting a constant level of volatility rather than a constant notional exposure: you lever up in quiet periods and scale down in turbulent ones.
The approach works because volatility clusters. Man Group notes that high volatility in the recent past tends to be followed by high volatility in the near future, an idea that sits behind Robert Engle's ARCH work, which earned the 2003 Nobel Prize in Economics. Volatility is much easier to forecast than returns, so a rule built on it has something solid to stand on.
One side effect is worth knowing. Man Group points out that left-tail events tend to be less severe, because they usually arrive when volatility is already elevated and your exposure has already been trimmed.
The formula, step by step
The calculation has three parts.
- Pick a target. Choose an annualized volatility you can live with, for example 15%.
- Measure or forecast volatility. Use a rolling standard deviation of daily returns, such as a 20-day window, or an exponentially weighted moving average (EWMA). A nullalpha.dev walkthrough lays out both options.
- Divide. Exposure = target volatility / forecast volatility. Cap the result at a maximum leverage and floor it at zero.
A worked example
Take a 10% target. Northstar Capital's example runs like this: if the asset is running at 20% volatility, you hold half a unit of exposure. If it calms to 8%, you hold one and a quarter units, subject to your cap.
Now apply it to a $10,000 account. At 20% asset volatility, you deploy $5,000 of exposure. At 8%, you deploy $12,500, which needs leverage of 1.25x. That is why the cap matters.
Why the cap is not optional
Quiet markets produce the largest positions. The nullalpha.dev write-up shows how an asset with 2% volatility and a 15% target would call for 7.5x leverage. The arXiv paper on smoothing volatility targeting found that for more than 10% of 158 equity strategies, simple realized-variance targeting required leverage between 1.8 and 4 times. Pick a hard ceiling before you start, and treat it as a safety rule rather than a tuning knob.
Choosing a volatility estimate
Your estimate drives everything, so the choice matters.
| Method | How it works | Main weakness |
|---|---|---|
| 20-day rolling standard deviation | Equal weight on the last 20 daily returns | A single shock spikes the estimate, then drops out about 21 days later |
| EWMA | Recent days weigh more, older days decay | Lags during sudden gaps |
| GARCH | Models volatility clustering directly | Only as good as the latest read, and more complex |
| ATR-based | Uses average true range as a risk unit | Not annualized, so it needs conversion for portfolio targets |
The rolling window's step-response problem is described in the nullalpha.dev post: the size jumps back up when the shock leaves the window, even if the market has changed for good. EWMA smooths that, but it can still understate risk for several days after a gap-down such as March 2020.
If you trade single positions with stops rather than a whole portfolio, our guide to ATR stop loss multipliers and sizing shows a practical way to put volatility into your stop distance and size.
What the research says about benefits
The case for volatility targeting is strongest on risk, less so on return.
- Man Group reports that targeting improves the Sharpe ratio of risk assets such as equities and credit, reduces the likelihood of extreme returns across more than 60 assets, and cuts maximum drawdowns for balanced and risk parity portfolios.
- Bongaerts, Kang and van Dijk proposed a conditional version that only adjusts exposure in extreme high- and low-volatility states. It consistently improved Sharpe ratios, reduced drawdowns and tail risk, and cut annual turnover to 1.6 from 2.4 in the US market. Leverage was capped at 200%.
- An Alpha Architect summary of the volatility-managed portfolio literature found the approach outperformed in 53 of 103 cases, with consistent Sharpe gains in only eight strategies, concentrated in momentum, profitability and betting-against-beta.
That last point is useful. Volatility targeting is not a free lunch for every strategy. It tends to pair well with trend-following, because volatility often rises as returns fall, and cutting size then points the same way a trend signal would.
What it looks like in crypto, forex and stocks
Crypto
Crypto volatility swings widely, which makes the method tempting. A backtest by Kryptera applied a GARCH(1,1) overlay to Bitcoin. Sharpe moved from 1.03 to 1.11 and max drawdown shrank from -62.24% to -17.52%, but CAGR fell from 41.64% to 14.30%. Read that as a trade: you buy a smoother ride and pay for it in raw growth. It is one author's backtest, so treat it as an illustration rather than a forecast.
Because crypto trades around the clock and gaps less often than stocks, many traders use daily returns with an EWMA estimate. Gaps still happen on exchange outages and liquidations, so keep the leverage cap tight.
If you manage BTC and ETH positions as longer-term holdings, this guide to sizing BTC and ETH positions covers how exposure can be scaled to a target risk level.
Forex
Major currency pairs usually move less than crypto, so the formula often calls for leverage. Brokers cap leverage and prop firms add daily loss rules, so your cap is set for you. Where you use a prop account, the drawdown bands used by prop traders are a natural ceiling for how far down-scaling should go. For the underlying arithmetic of pip risk and lot size, see our forex position sizing formula.
Traders who run paired accounts across a broker and a prop firm need to watch exposure on both sides, and hedging and exposure-control software can help coordinate sizing and loss limits in that setup.
Stocks
Stock indexes can gap overnight, which is the single biggest enemy of this rule. In a cash account without margin, you can only reduce exposure below 100%, never raise it above, a limit pfolio academy points out. That means you capture the de-risking half of the idea but not the leverage half.
Where volatility targeting fails
Honest limits keep you out of trouble. Plan for each of these.
It reacts after the event. The forecast is built from past returns, so the first day of a spike is taken at the previous size. A sudden gap does its damage before the forecast moves.
It can be pro-cyclical. EC Assets notes that the rule sells into falling markets and buys into calm, rising ones. In a V-shaped reversal it can cut exposure near the low and miss the rebound. A slow decline with moderate volatility, as in 2000 to 2003, may not trigger much de-risking at all.
Results vary by market. Bongaerts and colleagues found that conventional targeting failed to consistently improve performance in global equity markets and raised maximum drawdowns in the UK, Canada, Australia and Hong Kong by 4 to 34.4 percent.
Costs eat small accounts. In a 6-year TiPortfolio backtest on a $10,000 balance, the volatility-targeted portfolio paid $4.32 in fees against $0.94 for the fixed-ratio one and finished worse on most metrics. The author suggests weekly or monthly rebalancing and a longer lookback to cut false signals.
Out-of-sample gains shrink. EC Assets also cites Cederburg and coauthors, who found that most reported gains for volatility-managed portfolios disappear in realistic out-of-sample tests.
The practical lesson from Northstar Capital is that the risk control is dependable, while the improvement in risk-adjusted return depends on the asset. Use it for the first, and treat the second as a bonus.
How to implement it without overtrading
- Set a rebalance rule. Resize only when the target exposure moves by a set band, say 20%, or on a fixed weekly schedule. This echoes the conditional approach that lowered turnover in the research above.
- Smooth the estimate. Use EWMA or a longer lookback so one shock does not whip your size around.
- Cap leverage. Set the cap first. Never let the formula override it.
- Keep your stops. Kryptera's backtest write-up reminds readers that stop losses and hard drawdown limits do work a volatility overlay cannot. Our guide to stop loss placement at 1 to 2% risk shows how.
- Combine with a per-trade risk limit. Volatility targeting sets how big the whole book is. A rule for risk per trade sets how much any single idea can lose.
- Track drawdowns. Know what a deep loss demands to recover, as explained in our drawdown analysis.
If you are new to sizing, start with our risk controls for new traders before layering a volatility rule on top.
Testing before you risk capital
Because published gains often shrink in realistic tests, run your version on data it has not seen. Include fees, slippage and the delay between signal and fill. Then run it live on paper and compare the paper P&L with the backtest. Look at drawdown and realized volatility first, not just return, since those are what the rule is built to control. Check how it behaved in 2020, in a slow bear market and in a sharp reversal. If the results hold across all three, you have something worth sizing up gradually.
Perspective
Volatility targeting is a good example of what we care about at Discipline AI: evidence over prediction. The rule does not claim to know where price goes. It only says how much risk you should carry given what the market is doing now.
That fits how we built the platform. Sentinel, our intelligence and research system, studies market regimes, volatility, liquidity events and historical trade outcomes across multiple cryptocurrencies and timeframes. It helps traders see when conditions favor action and when they call for caution or standing aside. We think the second half matters as much as the first.
We also believe in practicing before committing. Paper trading, position sizing and leverage tools, and automated journaling let you test how a sizing rule behaves, then review the results with AI Trade Autopsies. We do not guarantee profitable trades. We aim to give you clear, measurable evidence so that your risk decisions are deliberate.
Tony
Put your sizing rules into practice
Discipline AI brings market-condition analysis, risk planning tools with position sizing and leverage calculation, paper trading and performance analytics into one mobile app for iOS and Android. You can rehearse a volatility-based sizing plan on paper, then review how it performed. Pro is $8.99 per month or $79.99 per year, and there is a $199.99 one-off option. See the Pro plans for details.
If you want a structured path to better habits, The Disciplined Trader is available for $79 one-off.
FAQ
What is a good volatility target for trading?
There is no universal number. Pick the annualized volatility whose drawdowns you can stomach, then size the target to it. Many multi-asset examples use figures such as 10% or 15%. Crypto traders often choose a higher target because the asset itself runs far hotter, but a lower target means smaller swings.
Does volatility targeting increase returns?
Not reliably. The research shows more consistent gains in risk control than in return. Alpha Architect's summary found outperformance in 53 of 103 cases, and the Bitcoin backtest above showed a lower CAGR alongside smaller drawdowns.
Which volatility estimate should I use?
EWMA of daily returns is a common starting point because it reacts faster than a flat window but avoids the step-response jump. A longer lookback reduces false signals. Whatever you choose, apply a leverage cap and test it on data you did not use to build it.
Can I use volatility targeting without leverage?
Yes. In a long-only, unleveraged account you can only scale exposure down below 100%. You keep the de-risking benefit in turbulent markets but give up the extra exposure in calm ones.
Sources
- The Impact of Volatility Targeting, Man Group
- Conditional Volatility Targeting, Taylor & Francis
- Smoothing volatility targeting, arXiv 2212.07288
- The Performance of Volatility-Managed Portfolios, Alpha Architect
- Volatility Targeting: Scaling Exposure to Vol, EC Assets
- Volatility targeting, Northstar Capital
- Volatility targeting: sizing by realized vol, nullalpha.dev
- Target volatility strategies, pfolio academy
- Sizing Bitcoin and Nasdaq with a GARCH volatility overlay, Kryptera
- Volatility Targeting with TiPortfolio, Tim Bai