AlphaAssayHow we grade ourselves — in public
Every rating business faces the same temptation: grade generously, keep customers happy, quietly forget the misses. We removed the temptation instead of promising to resist it.
Every verdict is a bet against the future
When we say fail, we are predicting that the signal would have disappointed on unseen data. When we say pass, we are sticking our neck out the other way. Both claims become checkable as real market data arrives after the verdict. So we check them — every day, automatically, and we publish the result.
The mechanics, in plain language
The daily record. Each day the outcome of our past verdicts is written into a public record. The endpoint is free and needs no account:
curl https://api.alphaassay.com/v1/public/calibration
Why we can't cheat it. Two reasons. First, every entry is cryptographically signed when written — editing the past would break the signatures. Second, our verdicts are demote-only: no rule exists by which a grade improves after the fact. The record can embarrass us; it cannot flatter us.
What the numbers mean. Over time the record shows a simple thing: when we said fail, did signals actually go on to disappoint? When we said pass, did they hold up better than chance? Statisticians call the summary a calibration score — you can just read it as „were their neins and jas worth anything?"
Zero is an honest number
Today the counter is low — the ledger counts from day one, and we launched recently. We could have faked momentum with invented numbers; a validation service that lies about itself would be a punchline. So the record starts where every honest record starts: at the beginning, in public, with the pen already locked away.
The same philosophy runs through everything else we expose: signed certificates anyone can verify, a public digest of failed strategy families, and timestamps that make rewriting history impossible rather than forbidden. Don't trust an examiner who won't be examined — examine us.