INTEL DESK · THESIS LEDGER

Public Scorecard

Every thesis on the dashboard is a falsifiable claim with a horizon. When the horizon elapses we grade it hit / miss / inconclusive and write the result to this page. We do not hide the misses. A thesis is currently graded on whether the desk’s own confidence resolved high or low at the horizon; grading against independent price and event outcomes is being wired in, and until then these figures measure self-consistency, not accuracy against reality.

Live queue 3 open claims Open theses stay visible before they can improve the public hit rate. Calibration gate Live grading · first close 23 Oct Closed 7d / 14d / 30d outcomes are the only denominator. Miss ledger No closed misses The highest-confidence misses remain attached to the proof page.

Hit rate

AWAITING CLOSE

0 hits · 0 misses · 0 inconclusive

Brier score

AWAITING CLOSE

Lower is better. Grading window opens at 60: 0 of 60 closed calls graded so far. Graded on confidence resolution, not yet against independent outcomes.

Sample

live grading since 16 Oct

0 calls graded so far. Grading window opens at 60. No historical claims were available to backfill, so grading runs live from 16 Oct and the first horizon closes 23 Oct. Every hit, miss, Brier point, and calibration bar fills from real closed outcomes only, never a backfilled estimate. Open claims remain visible below.

Open horizons under audit

3 live claims being scored

7d horizon · 53% · closes 23 Oct

Hormuz tanker transit volume falls more than 20 percent over the next 7 days.

3 signals · BZ=F / CL=F / GC=F / ^VIX

14d horizon · 52% · closes 30 Oct

A direct kinetic exchange between Iran and Israel occurs within 14 days.

2 signals · BZ=F / GC=F / ^VIX / ^GSPC

30d horizon · 45% · closes 15 Nov

Ukrainian frontline shifts more than 5km in either direction over the next 30 days.

0 signals · BZ=F / GC=F

Calibration curve

Each row is a confidence decile. The bar shows the resolved hit rate among theses we closed in that bucket. While grading is confidence-derived, this curve reflects how the desk’s confidence resolved rather than a calibration against independent outcomes, so it cannot yet be read as a true reliability diagram.

Live calibration windowNo closed claims yet. Grading is running from 16 Oct; the curve will populate from the first settled horizon without inventing historical outcomes.

Biggest misses

No closed misses yet. Live grading began 16 Oct; every miss will remain listed here, ranked by confidence at close.

Generated 17 Oct 2026, 09:42 UTC Engine ticks every 5 minutes Hit rate denominator excludes inconclusive outcomes