Results

How the calls came in.

Every DeepStack story makes a call and names the tests that would prove it wrong. When a test resolves, the result lands here: the verdict, the evidence and its sources, and what it did to the call.

The scorecard

Counted in the open.

No accuracy rate is shown until ten results exist. None so far. Counts by verdict appear from the first result; the rate appears at the tenth, because a smaller sample flatters or punishes by chance.

Results
0
Confirmed
0
Refuted
0
Inconclusive
0

Nothing has resolved yet

The first result is on the clock.

Next up: Oct 15, 2026. Treasury publishes August foreign holdings; renewed Japanese buying despite poor hedged returns would undercut the missing-buyer thesis behind the long-end selloff. Part 07.

  1. 01

    The verdict

    Confirmed, refuted or inconclusive, decided against the test the story wrote before the event, never after it.

  2. 02

    The evidence

    The figures that settled it, each with its source and its date, so the result can be checked.

  3. 03

    The call

    What the result did to the DeepStack call: conviction up, down or unchanged, and why.

  4. 04

    The record

    Counts by verdict from the first result. No accuracy rate until ten exist.

Open the Docket Add the dated tests to a calendar

Results are published only after a human review. Analysis and opinion, not investment advice.