Position sizing under crypto volatility

Definition

Crypto position sizing is the standard sizing tree (risk-fixed-fractional → risk-volatility-targeting → drawdown budgets) run with crypto-grade inputs — and the arithmetic is unforgiving: at realized volatility running at multiples of equity levels (Liu-Tsyvinski 2021; crypto-volatility-character), any honest risk budget produces allocations that look SMALL next to the asset’s narrative footprint. The entry’s one theorem: if the sizing answer for a crypto position resembles the sizing answer for an equity position, one of the inputs is wrong. Structural loss modes (crypto-loss-modes) then stack a second budget — venue and custody exposure — on top of price risk.

How it works / structure

• Vol-normalized base: volatility targeting sizes positions inversely to measured vol — a book targeting 10% annualized portfolio vol holds a 3-5x-equity-vol asset at a fraction of an equity weight mechanically (risk-volatility-targeting); ATR- or realized-vol-scaled units carry across from futures practice.
• Tail-scaled stops and budgets: per-trade risk (risk-fixed-fractional) uses stop distances that respect crypto’s documented daily tails — stops inside the asset’s routine daily range are churn, not protection (crypto-volatility-character); drawdown budgets price the full-history record, where 70%+ asset-level excursions recurred (crypto-drawdown-behavior).
• Kelly discipline: with fat two-sided tails and unstable parameters, Kelly fractions computed from trailing samples overstate — the Kelly-as-ceiling rule (risk-kelly-criterion) binds hardest exactly here.
• The custody budget: exposure AT a venue is a position in the venue (crypto-custody-models); sizing caps per venue and per custody mode sit alongside the market-risk budget — a second constraint, not a substitute.
• Leverage interaction: leveraged crypto structures (futures margin, perp leverage) compound vol-of-vol with liquidation mechanics — effective leverage caps below venue maxima are structural, not conservative (crypto-perpetual-futures).

When it applies

Every crypto position, without exception — and the frame binds tightest for: additions to an existing equity book (the correlation input is the stressed value — crypto-correlation-regimes), income structures on crypto underlyings (short-premium sizing against crypto tails — crypto-options), and venue-resident strategies where working capital carries custody exposure continuously.

Risk profile & failure modes

• Narrative sizing: conviction-scaled rather than vol-scaled positions — the equity-habit error crypto punishes fastest; the documented tail record is the input, not the thesis’s confidence (bias-overconfidence).
• Calm-regime creep: compressed-vol eras invite size creep that the next regime shift marks down at crypto amplitude (bias-recency, regime-volatility).
• Single-budget accounting: books that price market risk but not venue concentration discover the custody budget at failure time (crypto-loss-modes).
• Stop-distance denial: tight stops on a fat-tailed 24/7 asset generate weekend gap fills far through the level (crypto-sessions-24-7) — the realized loss exceeds the budgeted loss by construction.

Evidence & limits

Volatility magnitudes are peer-reviewed; drawdown history is public record; the sizing frameworks are the platform’s documented tree with their own evidence bases. No target allocation, leverage number, or per-venue cap is prescribed — those are outputs of each book’s budget arithmetic, produced by the frameworks with current measured inputs.

Falsifiable-thesis examples

Illustrations only, not signals:

• “A vol-targeted crypto sleeve (10% portfolio vol budget) realizes portfolio drawdown under 15% through the next crypto drawdown exceeding 40% (budget-integrity thesis)” — falsified by the realized path.
• “Fixed-fractional sizing with ATR-scaled stops produces smaller realized-vs-budgeted loss gaps than fixed-percent stops over the replay sample (stop-calibration thesis)” — falsified by the paired replay.

Cross-references

• The sizing tree: risk-fixed-fractional, risk-volatility-targeting, risk-kelly-criterion, risk-max-drawdown-budget
• The inputs: crypto-volatility-character, crypto-drawdown-behavior, crypto-correlation-regimes
• The second budget: crypto-custody-models, crypto-loss-modes