The scorecard
Every committed call, graded against the market baseline — Brier, log-loss and CLV, with the calibration curve. Losers included. This is the deep layer behind the live match slate.
Plumb v1 does not trade and is not sold. "Evidence-backed" is honestly partial today (LLM reasoning + smart-money and news priors). "Powered by the τ engine" is provenance only.
Early days — only a handful of calls have resolved, so the record is still building. The numbers below are directional until the sample grows.
Plumb vs the market
Not enough resolved calls to compare Plumb against the market baseline yet.
Brier lower is better
— Plumb
— Market
Log-loss lower is better
—Plumb
—Market
Mean CLV vs closing, higher is better
—Plumb
—n/a for market
Low sample: too few graded calls for a stable read — the record is still building. Treat these numbers as directional, not conclusive.
Calibration
Not enough resolved calls yet to plot calibration — the record is still building. Points appear here as graded predictions accumulate across the probability range.
Full call log
Every committed prediction — including losers, abstains, voids and non-scoring calls. Nothing is hidden. Each row links to its on-chain audit.
| Date / market | Side | P(YES) | Mkt price | Outcome | Brier P / mkt | CLV | Status |
|---|---|---|---|---|---|---|---|
| 2026-06-28 0x31cf97…afd7 | YES | 0.27 | 0.27 | — | — / — | — | Pending |
| 2026-06-28 0x355ecd…82b7 | YES | 0.17 | 0.17 | — | — / — | — | Pending |
| 2026-06-28 0x688954…32ed | NO | 0.56 | 0.57 | — | — / — | — | Pending |
| 2026-06-28 0x84466f…940b | NO | 0.25 | 0.27 | — | — / — | — | Pending |