Public methodology mirror for Outlier Weekly — three-formula synthesis (Poisson, Shannon entropy, Taleb fat-tail) applied to rare-event prediction markets on Polymarket and Kalshi.
View the Project on GitHub cemini23/outlier-weekly-methodology
Every call made in Outlier Weekly is logged on this page the day it ships, with the entry price, the sized fraction, the resolution date, and the eventual outcome. Track-record discipline begins with Issue 1.
Rules of this sheet:
| Issue | Ship date | Market | Outcome window | Direction | Entry (YES¢) | Sized fraction | Resolution date | Status | Resolution | Issue link |
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 2026-05-14 | Will Anthropic make a deal with the Pentagon by [date]? | June 30, 2026 | SKIP | 32¢ (draft) → 27¢ (publish-day lock) | 0% of bankroll | 2026-06-30 | OPEN | pending | Issue 1 |
Market: Will Anthropic make a deal with the Pentagon by June 30, 2026? Direction: SKIP — model–market convergence inside the noise floor.
Model output (draft, p_market = 32¢):
Model output (publish-day, p_market = 27¢):
Why the change between draft and publish: the market moved 5pp between the analytical work and the publish-day lock, into the immediate neighborhood of the Poisson fair value. The directional signal that read BUY_NO at 32¢ flattens to a near-zero edge at 27¢. Under the synthesis’s skip-discipline rules, this collapses the weighted output to an effective non-position. The methodology refuses the trade when market and model converge inside the noise floor — Issue 1’s forward-track row is the first published instance of that refusal.
Resolution oracle: Polymarket UMA / public announcement of qualifying agreement, evaluated against the market’s rules text as of 2026-05-14.
Re-watch trigger: If the June-30 price moves outside a 22–32¢ band before resolution, the synthesis is re-run and any resulting call is logged in the next issue.
Future entries will follow this column ordering. Definitions:
When SKIPs resolve, they are evaluated as either SKIP-CORRECT (the position the methodology would have taken would have lost, so the refusal was right) or SKIP-WRONG (the position would have won, so the refusal cost the portfolio). Track-record discipline is measured against both kinds of calls.