Finality Rules
How signals become verdicts. Why most proposals never reach finality.
What is a Signal?
A signal is a proposal generated by King intelligence. It represents a directional hypothesis derived from on-chain and market features.
Signals contain:
- •Direction — LONG, SHORT, or NEUTRAL
- •Confidence — Model certainty (0-100%)
- •Features — On-chain inputs (fees, mempool, funding, hashrate)
- •Thesis — Natural language reasoning
A signal alone authorizes nothing. It is merely a candidate for validation.
What is a Verdict?
A verdict is a finalized decision that has survived the full validation pipeline. Only verdicts authorize capital deployment.
A verdict is earned when:
- ✓Swarm validators achieve consensus on structural integrity
- ✓Miners expend compute attempting to break the signal
- ✓Signal survives adversarial perturbation within bounds
- ✓Minimum work threshold is met
Capital does not move on signals. It moves on verdicts.
How Finality is Earned
Stage 1: Proposal
King generates signal envelope with direction, confidence, features, and thesis. Signal enters UNCONFIRMED state.
Stage 2: Swarm Vetting
Swarm validators independently evaluate signal structure. Each agent votes CONFIRM, DISSENT, or ABSTAIN with reasoning. Integrity score calculated. High dissent flags structural weakness.
Stage 3: Miner Attack
Miners expend compute attempting to find counterexamples. Witness proofs demonstrate robustness under perturbation. Counterexample proofs identify minimal deltas that flip the signal.
Stage 4: Finality Decision
Based on accumulated evidence, verdict is issued:
- ● HARDENED— Signal survived, ready for execution
- ★ REWARDED— Position closed profitably, rewards distributed
- ○ REJECTED— Signal failed validation, no execution
State Machine
┌─────────────────────────────────────────────────────────┐
│ SIGNAL LIFECYCLE │
└─────────────────────────────────────────────────────────┘
┌──────────────┐
│ PROPOSED │ ← King generates signal
└──────┬───────┘
│
▼
┌──────────────┐
│ UNCONFIRMED │ ← Awaiting swarm vetting
└──────┬───────┘
│
▼
┌──────────────┐
│ VETTED │ ← Swarm approved, miners working
└──────┬───────┘
│
├─────────────────┬─────────────────┐
│ │ │
▼ ▼ ▼
┌──────────────┐ ┌──────────────┐ ┌──────────────┐
│ HARDENED │ │ DEGRADED │ │ REJECTED │
│ (execute) │ │ (weak conf) │ │ (no execute) │
└──────┬───────┘ └──────────────┘ └──────────────┘
│
▼
┌──────────────┐
│ RESOLVED │ ← Position closed
└──────┬───────┘
│
▼
┌──────────────┐
│ REWARDED │ ← Profits distributed
└──────────────┘
TRANSITIONS:
────────────────────────────────────────────────────────
PROPOSED → UNCONFIRMED : Signal envelope complete
UNCONFIRMED → VETTED : Swarm integrity > threshold
VETTED → HARDENED : Miners confirm robustness
VETTED → DEGRADED : Weak but not fatal
VETTED → REJECTED : Multiple counterexamples
HARDENED → RESOLVED : Position closed
RESOLVED → REWARDED : PnL > 0, rewards distributed
Why Most Signals Never Become Verdicts
The validation pipeline is intentionally adversarial. Most proposals fail before reaching finality.
When validator agents disagree, structural weakness is indicated. Signals with >40% dissent rarely achieve finality.
Miners finding small perturbations that flip the signal indicate fragility. Low-distance counterexamples trigger rejection.
Minimum compute threshold must be met. Signals cannot harden without adequate miner participation.
Witness proofs with poor pass rates indicate signal instability under perturbation. Consensus is required.
Why Speed is Secondary to Conviction
This system does not optimize for signal frequency or speed. It optimizes for decision quality.
A single high-conviction verdict that survives adversarial testing is worth more than many fast, untested signals.
Traders who need speed over conviction should look elsewhere. This protocol exists to produce verdicts worth trusting.
Signals are proposals.
Verdicts are earned.