Five questions. Two answers are no.
Anyone can publish the result that flattered them. The only cheap way to tell careful measurement from measurement that stopped when it liked the number is to publish the refusals too — with the arithmetic attached, so you do not have to believe this paragraph. Every figure below is read from a JSON artefact written by a command in the repository.
Reading the artefacts…
Reproduce any of it
Each artefact is regenerated by one command, and each command reads only public endpoints — no key, no funded wallet, no access to this deployment.
| Question | Command | Writes |
|---|---|---|
| 00 · one clean order | npm run final-proof | final-proof.json |
| 01 · on-chain | npm run proof | live-canary.json |
| 02 · the model | npm run calibrate | calibration.json |
| 03 · portfolio layer | npm run backtest | backtest.json |
| 04 · making | npm run maker:live -- --live | maker-live.json |
| 05 · coherence | npm run coherence -- --days 30 | coherence.json |