What doesn't the ledger prove?

It proves what was logged, when, and how it was scored. It doesn't prove that nothing was left out, that a writer is skilled, or that the prices used to score a call were correct.

Details

• Omission. A writer can choose not to log a call. Profiles say "calls logged by the forecaster" and can link each call to its original post.
• Publish time. For self-logged calls, the time a writer says they first published is self-reported. It doesn't affect scoring (the window can't open before the call is recorded), and it isn't presented as a verified publish time.
• Scoring inputs. Verification recomputes each score from the recorded inputs, but doesn't check those inputs against market data. Pinning the raw price files by hash is part of the design, and independent spot-checks are planned.
• Independence of calls. Many calls on one idea, or on stocks that move together, look like more evidence than they are. Grouping correlated calls is required before any public ranking.
• Skill or advice quality. We verify timing and outcome, never the quality of anyone's advice.
• Garbage in. A wrong price or a misread post is recorded faithfully. Corrections are visible records.