the market data source that tells you what it doesn't know
GLOOMBORG issues a verdict on a market claim and shows its working — whether the verdict is flattering or not. Not "here is data". Not "here is analysis". A tested answer to a question someone actually believes.
| The claim people believe | GLOOMBORG's verdict |
|---|---|
| Boycotts move share prices | Median −3.1% over 20 days. Yes, modestly.event study · 20-day excess return |
| Companies go quiet before bad news | No. No detectable lead.pooled lead correlation r=+0.015 · p=0.83 · n=203 |
| Our own trading signal has an edge | No. No edge in-sample.backtest · published, not hidden |
| The January barometer predicts the year | No — noise.folklore test · Halloween holds · January fails |
| "Record / worst / unprecedented" means something | Hype words correlate with clicks, not accuracy.language frequency vs outcome |
Most of those are null results. They are not failures of the product — they are the product. A source that reports only positives is a source nobody believes.
Every number carries its method, its coverage, its licence, and its clock-drift. "Nobody publishes their own losing backtests" — we do.
"No detectable lead" is what makes our positives credible. We tell you when there's no signal — which is why you believe us when there is.
We archive public records, not rent other people's APIs. It cannot be relicensed out from under you.
A report hands you a conclusion and keeps the evidence. We hand you the evidence and keep nothing.
| Refused | Why |
|---|---|
| Not a Bloomberg clone | Indexes and price feeds are licensed products we cannot own and do not need. |
| Not a signal seller | We publish our own nulls. Selling signals requires the opposite. |
| Not a conclusion machine | We deal the cards; we do not tell you what to think. |
| Not an accuser | We measure footprints. We do not allege intent. |
| Not advice | This is the boundary that makes the rest possible. |