By Keel Research Team · Updated May 13, 2026
Volatility targeting is a position-sizing technique that scales exposure inversely to recent realized volatility — bigger positions in calm regimes, smaller in volatile ones. The result is steadier portfolio risk across regimes, better Sharpe, and easier-to-hold drawdowns at the cost of leaving some upside on the table in strong trends.
Most position sizing rules are static — fixed % of account per trade, or a fixed leverage. They work fine in stable regimes but produce lumpy realized risk: in high-vol periods, the same dollar position carries more risk than in calm periods. Drawdowns and run-ups both get amplified.
Volatility targeting fixes this. Instead of fixing position size and letting risk fluctuate, it fixes risk and lets position size fluctuate. The intuition: take the same amount of risk in every regime; the market is more violent some weeks than others, so adjust accordingly.
Pick a target portfolio volatility — typical values: 10-20% annualized for systematic trend strategies, 5-10% for institutional-grade. Then:
position_scale = target_vol / current_realized_volIf your portfolio currently realizes 15% annualized vol and you target 15%, position scale = 1 (full position). If current vol jumps to 30%, position scale = 0.5 (half position). If current vol drops to 7.5%, position scale = 2 (double position).
Realized volatility is typically computed as a rolling N-day standard deviation of returns, then annualized. Common lookbacks: 20 days for crypto (faster response to regime change), 60 days for traditional assets (smoother).
At the asset level, the same logic applies using each asset's own ATR or stddev — inverse-ATR sizing makes a low-vol BTC position bigger than a high-vol DOGE position for the same risk budget.
Sharpe ratio = return / vol. Holding returns roughly constant while reducing the variance of realized vol mechanically improves Sharpe — that's the headline.
The deeper reason: volatility is autocorrelated. High-vol periods cluster together; low-vol periods do the same. A static position size systematically over-risks in high-vol regimes (where the next move is also likely to be large) and under-risks in low-vol regimes (where the next move is likely to be moderate). Vol-targeting counters this autocorrelation, reducing both upside and downside variance in the realized return distribution.
Concrete effect: a trend strategy with static sizing might have Sharpe 1.0 over a multi-year sample. Adding vol-targeting often pushes that to Sharpe 1.3-1.5 — same edge, smoother experience.
Vol-targeting is not free. Two regimes where it costs you absolute return:
Most trend systems use vol-targeting anyway because the path-pain reduction is worth the absolute-return cost. For Sharpe-maximizing investors, vol-targeting is almost always net positive. For pure-upside-seeking traders willing to accept volatile equity curves, static sizing in strong-trend regimes can outperform.
Vol-targeting works as a portfolio-level layer on top of per-trade rules:
The combination gives per-trade loss containment, edge-aware aggression, and portfolio-level regime smoothing.
The Keel pipeline includes vol-targeting as a built-in component. Add it to your strategy and configure the target vol + lookback window. The trend-following template uses this pattern — position sizes inversely to volatility, portfolio rebalances hourly. Open the template, inspect the pipeline, fork into your own workspace to modify.
Keel is a Strategy OS for AI-assisted systematic trading on Hyperliquid. Backtest, optimize, and run live strategies across single-stock perps, indices, and crypto majors — realistic fees, slippage, and funding modeled.
Free to start — connect a Hyperliquid wallet when you’re ready to go live.
A sizing rule that adjusts position size inversely to recent realized volatility. Bigger positions when an asset (or strategy) is quiet, smaller when volatile. The goal is to keep total portfolio volatility steady regardless of market regime, rather than letting realized risk swing wildly with the market.
Three reasons. (1) Better Sharpe ratio — keeping vol steady tightens the return distribution. (2) Smoother equity curve — drawdowns shallow in high-vol periods because positions are smaller. (3) Better psychological survivability — consistent risk profile is much easier to hold through than lumpy returns.
Strong directional regimes. When a trend is running and vol is rising, vol-targeting cuts position size — exactly when you'd want to be bigger. So the strategy leaves upside on the table in strong-trend environments. Trade-off for the smoother ride.
Pick a target portfolio vol (e.g. 15% annualized). Compute realized vol on a rolling window (e.g. 20-day standard deviation, annualized). Set position scale = target_vol / current_vol. When current vol matches target, full position. When current vol is double target, half position. Etc.
Risk-percent fixes loss-per-trade as a constant. Vol-targeting fixes portfolio vol as a constant. They operate at different layers — you can use both: risk-percent per-trade for loss containment, vol-target at portfolio level for regime smoothing. Most production trend strategies combine them.
Yes. The Keel component system includes vol-target sizing as a built-in primitive. Strategies that include a vol-target stage automatically scale positions based on realized vol of the universe or the strategy's own returns. See the trend-following template for a worked example.