Ex-post best-execution audit
Checking every fill against a public benchmark after settlement.
Once a trade settles, its input amount and route are public. For each settlement, subnet miners:
| Step | Provable / public | Hidden |
|---|---|---|
| 1 · Reconstruct | Pool states recorded at the settlement block | — |
| 2 · Benchmark | Maximal achievable output for the revealed amount | — |
| 3 · Compare | Delivered output vs. benchmark | — |
| 4 · Score | Flag if deviation exceeds the tolerance band; update integrity score | — |
The tolerance band allows for normal latency and slippage. Validators combine miner results through Yuma consensus and write them to the Solver Registry. A confirmed failure puts the solver's $OCLU bond at risk (see Solver bonding & $OCLU).
The audit is cheap because it runs on public data after the fact. The trade-off is a short, limited window before settlement in which execution has not been audited yet.