Skip to main content
All topics

Risk Management

How position sizing, vol-targeting, and multi-layer protection works.

What the orchestrator manages automatically

You don't configure these — they're baked into the engine: - Vol-targeting — daily realized vol target 1.6%, capped at 100% gross, floored at 30%. - Greedy decorrelation — top-5 sleeves, max pairwise correlation 0.7. - Per-class DD-brake (Phase 82) — trend 15%/5%/40%, carry 8%/3%/40%, xcarry 6%/2%/40%. - Global DD breaker — 20% peak-to-trough halt with admin notification. - Capital safeguards — cold-start cap, daily P&L breaker, order anomaly detection. - Intraday VaR monitoring — every 5 minutes with warn/crit alerts. - Per-exchange circuit breaker — repeated errors trip the exchange off for 60s. - Kill-switch — 100ms TTL cache, checked per-order (not just per-tick). - Pre-trade limits — configurable per risk level.

What YOU control

Only two primary knobs: - Risk Level — Conservative / Balanced / Aggressive. This sets all downstream parameters. - Allocation $ — how much of your exchange balance Smartbull is allowed to deploy.

That's it. There's no "stop-loss tightness", no "indicator length", no "leverage slider". If you want to dial down risk, switch to Conservative or lower Allocation $. Halving allocation halves every position.

Advanced users (Alpha+ tiers) can also toggle: - AI Autopilot vs Proven strategies mode - Adaptive allocation (Meta-Learner on/off) - AI overlay (ensemble size multiplier on/off)

Paper-trade first

The bot ships with paper mode ON. In paper mode we evaluate every tick, log the planned orders, and simulate fills with realistic 10bps slippage — but never call the exchange.

You should leave paper mode on for at least 5-7 days to observe a full regime cycle before going live.

To flip live, toggle Paper mode off in /settings → Bot. You'll get a one-time risk-disclosure modal.