These are paper leans, not model-recommended BETs. They narrowly miss the numerical point-edge requirement, but the full audit is still incomplete. See the exact price, why each fell short and how the original observation performs.
On the radar
A close number is a research question. It does not establish a profitable bet. Captured prices expire after 20 minutes; missing evidence is stated explicitly.
The paper trail
A separate prospective record. We save the first qualifying fresh observation before kickoff, preserve its line and price, and retain losses and pending games. Previews using old prices do not count.
What gets tracked?
The estimated point edge must miss the BET requirement by no more than 0.5 points, while clearing the same numerical expected-return and win-probability minima. Only valid spread or total calculations enter. The source quote must be fresh when the paper record is created. These estimates are unvalidated; personnel and adverse-scenario audits remain incomplete.
The first entry for each game, market and direction is retained across books and later prices. A refresh cannot rewrite an earlier lean. Multiple markets in the same game are correlated observations.
Fully audited near misses
A different cohort: completed audits that narrowly miss the formal BET gates, including their stress checks. Those stricter rules and results remain intact.
Paper results assume one hypothetical unit at the original captured price. Final scores need reviewed source evidence; unresolved and postponed games stay pending. Hypothetical returns are separate from official BET performance and actual wagers. A winning streak does not by itself justify loosening the model.
