Coordinating agent execution. Validate, size, and audit every order against your portfolio's risk profile.
For funds, prop desks, and teams building trading agents.
Guardrail sits in the order path. Every proposed trade is checked, sized, and recorded against a client-specific risk profile before it reaches the market.
Run multiple strategies or agents in parallel. Guardrail evaluates competing proposals against live portfolio state and a single risk profile, then returns the approved action and size—including which constraints bound and why.
// Competing proposals (same portfolio)
POST /v1/guard/decide
{
"portfolio_id": "acct_0042",
"profile_version": "risk_profile:v3.1",
"objective": "maximize_expected_utility",
"proposals": [
{
"agent": "Momentum-A",
"symbol": "ES.F",
"size": 5.0
},
{
"agent": "MeanRev-B",
"symbol": "ES.F",
"size": 4.0
}
]
}
// Decision
{
"decision_id": "dec_01HT…",
"selected": {
"agent": "MeanRev-B",
"status": "MODIFIED",
"approved_size": 3.2
},
"binding_constraints": ["CVAR_99", "LIQUIDITY"]
}Define a versioned risk profile (limits, exposures, drawdown tolerance, restricted assets). Guardrail compiles it into enforceable logic and applies it at order time—using scenario-based optimization and stress checks to approve, resize, hedge, or block trades—with a clear record of what was binding.
// Incoming proposal
POST /v1/guard/check
{
"portfolio_id": "acct_0042",
"profile_version": "risk_profile:v3.1",
"scenario_set": "base+stress",
"proposal": {
"symbol": "ES.F",
"size": 5.0,
"intent": "aggressive_breakout"
}
}
// Guardrail decision (12ms)
{
"decision_id": "dec_01HT…",
"status": "MODIFIED",
"binding_constraints": ["CVAR_99", "LIQUIDITY"],
"action": {
"approved_size": 3.2,
"hedge_required": "ZN.F SHORT"
},
"audit_record": {
"portfolio_snapshot": "ps_8f2c…",
"profile_hash": "0x91c4…"
}
}