Every price below buys a complete, signed, replayable verdict — not a quota, not a seat. Agents pay per call; nobody pays to find out whether we're any good — the free tier exists exactly so you can test us first.
| product | what you get | price | notes |
|---|---|---|---|
| demo / golden specimens | known-answer test cases, full response format | $0 | rate-limited, no account needed |
| calibration ledger | daily public record of how our verdicts turned out | $0 | public API, no account needed |
| certificate verification | authenticity check of any AlphaAssay verdict | $0 | on /verify or offline |
| validate — signal | full battery + verdict + failure codes + signature | $0.50 | first signal of a family |
| validate — variant | same, for another variant of the same family | $0.10 | tweaked parameters, same idea |
| validate — backtest | overfitting forensics on backtest results | $0.50 | equity curve, returns or trade list |
| pre-register | tamper-proof public timestamp for a hypothesis | $0.10 | seal now, evaluate later |
| forward evaluation | strictly post-timestamp scoring of a registration | $0.25 | per evaluation run |
| forensics report | full autopsy: decay, costs, regimes, capacity | $2–5 | depends on data volume |
| signed certificate | exportable, independently verifiable document | $29–49 | per certificate |
AI agents pay per request via the x402 protocol — machine-to-machine payment, no API-key ceremony, no account. The price is quoted in the response before anything is charged.
Running a desk, a fund or serious volume? We set you up personally — terms, invoicing, limits. Write to hello@alphaassay.com.
A verdict is priced when it completes. Rate limits protect the free tier; paid calls are never throttled below what you paid for.
| Why do variants cost less? Because testing your fifth parameter-tweak is statistically cheaper than opening a new strategy family — and because honest multiple-testing accounting means the family shares one budget. We price the way the statistics work. |
| What if the verdict is „fail"? You pay the same. You are buying the trial, not a flattering outcome — that is the whole point. Every fail comes with the diagnosis of what broke, which is usually worth more than a pass. |
| Can I verify I got what I paid for? Yes — every verdict is signed and bit-identically replayable. Check it on /verify or offline with our published key. |
| Does a „pass" mean I will make money? No, and we will never claim that. A pass means your signal survived a statistical trial that most signals fail. Markets change; methodology audits are not investment advice. |
AlphaAssay