Volatility

Volatility quantifies the dispersion of returns, without regard to direction:

σ = std(returns over a lookback window)
annualized σ = σ per period × √(periods per year)

Daily returns with a 2.5% standard deviation annualize to roughly 2.5% × √365 ≈ 48%, a typical order of magnitude for BTC. Equity indices run closer to 15 or 20%.

Why strategies must adapt to it

Crypto volatility is strongly regime-dependent. Weeks of 1% days give way to 8% days without notice, and fixed-width stops and fixed position sizes built in one regime malfunction in the next: stops get noise-hit in high vol, and sizing risks too much of the account. The standard fix is volatility scaling, meaning ATR-based stops and vol-targeted position sizing, which keep risk per trade roughly constant across regimes.

Volatility as signal, not just risk

Some strategies trade volatility itself. Breakout systems demand a range expansion before entering (the "volatility expansion" filter in a Donchian breakout), and squeeze setups on Bollinger Bands bet that unusually quiet markets precede violent moves.

On AlphaProve

The vol-scaling the section above describes is built in on two fronts: the adaptive helpers volatility_target_risk and atr_risk turn a volatility estimate into a risk fraction, and the risk_per_trade sizing mode then holds dollar risk per trade roughly constant as ATR widens and narrows. When a run spans a regime change, the tearsheet's price-shock attribution card isolates how much of the result came from a handful of high-volatility bars, the usual place fixed-size strategies make or lose their year.