Can ZEME prove what it said before the market moved?
Every number on this page is computed from predictions that were sealed with SHA-256 before their outcome existed, then scored against the public tape by stated rules. Nothing here is a marketing claim: every row links to a chain record you can re-hash in your own browser, and every outcome is recomputable from the frozen fields. Misses included — they are the point.
Close-to-close from the frozen entry, across every scored pick — winners and losers together. Median first: it is the number a single pick most resembles.
ZEME shows a confidence figure beside its signals. This table is what each band has historically delivered, measured on every resolved signal in the internal ledger — a larger population than the sealed set above, and a different one: these rows are not tamper-evident, and this scale is not the stock-page confidence number. Two rates are shown because both denominators are honest: decided (wins vs losses) and all resolved (expiries included).
The phase each pick was sealed in. Groups with fewer than 10 decided picks show their count but no percentage — a rate from a handful of picks is a coin story.
Regime on the pick date: KSE-100 above (UP) or below (DOWN) its trailing 50-session mean.
The full record — each row carries its hash prefix; verify any of them on the freeze ledger.
Records that cannot be honestly scored, each with its reason. Dropping them silently would flatter every number above.