Briefs
Analysis notes built on the finished analytics, each dated to the vintage it was true of. A brief here does not type its numbers: every figure is queried from the parquet estate when the page is built, so if a model is refitted, the prose moves with it. A hand-typed figure is a claim. A queried one is a measurement.
- 6 min
The one series that saw the panic is published two years late
In the quarter Silicon Valley Bank failed, banks drew more from the Fed's discount window than in the whole of the rest of the disclosed record put together. By law, nobody outside the Fed could see it for two years. That is not a scandal. It is the price of the window working at all, and it puts a hard ceiling on what any early-warning system can do.
- 7 min
A banking system does not lose half its capital in a profitable year
In 2024 Bangladesh's banks reported a profit, and reported capital adequacy falling from 13.15% to 5.59%. Both cannot describe the same year. The arithmetic says what happened: the losses were already there, and the accounts had not said so.
- 6 min
Every model here is published beside the thing that beats it
FinObservatory runs three predictive models. Each one is scored against the simplest rival that could embarrass it: one ratio, or one variable. Two of the three lose outright. The third wins by so little, against a rival so crude, that the conclusion does not change.