LucraX ships seven position-sizing methods. They differ in two dimensions: do they adapt to volatility, and do they adapt to recent performance.
At a glance
| Method | Adapts to volatility | Adapts to performance | Best for |
|---|---|---|---|
| risk_based | Yes (via stop distance) | No | Most strategies |
| kelly | No | Yes (W/R from history) | Experienced + stable edge |
| half_kelly | No | Yes | Balanced Kelly |
| quarter_kelly | No | Yes | Conservative Kelly |
| volatility_adjusted | Yes (direct ATR) | No | Stops not tied to ATR |
| fixed | No | No | Testing |
| percentage | No | No | Buy-and-hold / spot |
Picks by strategy type
- Crypto trend follower with ATR stop → risk_based.
- Mean-reversion with band-based stops → volatility_adjusted.
- Established edge with 50+ trade history → half_kelly.
- Brand-new strategy under testing → fixed.
- Buy-and-hold accumulator → percentage.
Anti-patterns
- Full Kelly on a new strategy - drawdowns can hit 50% before the estimates settle.
- Fixed sizing in live trading - no risk control, scales poorly.
- risk_based with a 10% stop and 5% risk-per-trade - a single loss takes 5% of the account; 10 in a row would halve it. Tune risk per trade percent down or shorten the stop.
Stacking with risk management
All sizing methods respect max position percent and the user-level circuit breakers (daily loss limit, max drawdown). Sizing decides the trade; risk management decides whether the trade fires at all.