The Receipts
The Forecaster Scorecard
Everyone you follow makes predictions. Nobody checks them. We capture the calls of named investors, scientists, and macro thinkers — dated and sourced — and grade them when they mature.
What exists today
- 12,447 dated predictions captured from the knowledge base (as of Jun 29, 2026)
- 1,894 carry explicit time horizons — mechanically gradable as they mature
- 0 grades published so far. We will not publish a grade before the methodology is public and fixed.
The corpus is built from public statements — podcasts, interviews, filings — each stored with the speaker, the date, the source, and the claim in structured form. That is what makes grading possible at a scale no human fact-checker can match.
The grading rules, fixed in advance
Claims, never persons
We grade what was said, with the receipt attached — respectful, dated, sourced. Never a character judgment.
Verbatim or nothing
No claim is graded unless we can source it verbatim with a date.
Methodology published first
The grading rules are public before the first edition, and they do not move after the fact.
We grade ourselves
Synthos’s own calls — including the thinkers our system weights most heavily — are graded by the same rules, misses included.
Standing corrections policy
Any graded party can dispute a grade; disputes and corrections are published on the same page as the grade.
What is coming
“Who was actually right?” — quarterly, free
Every tracked forecaster’s horizon-bearing calls, graded against outcomes: hit rate, notable calls, notable misses, all dated. Editions are free to read — accountability shouldn’t sit behind a paywall.