Everything the model studied.
Including what it rejected.
The research desk behind the official board: strongest non-promoted candidates, probability and fair-price context, line movement, trend history, player props, game-level diagnostics and the exact evidence keeping each angle off the main model.
What Bosh actually knows right now
Live coverage and long-term market memory behind the research desk. Raw sportsbook observations remain short-lived audit evidence; researched consensus snapshots, movement history and recommendation closes are preserved for longitudinal analysis.
Best research that did not make the official board
Best bets that did not clear the main-model criteria. Ranked from the current research run. Each card shows the best available case, the strongest counter-case, probability context, market movement, the visible test battery and what would have to improve before promotion.
Where the secondary board is failing
This summarizes the visible evidence checks across today’s research-only candidates. It turns “did not qualify” into an inspectable reason instead of a black box.
Interactive game intelligence
Featured games with lead research, attached props and an expandable market/dependency lab. Use this as the one-screen game dashboard before drilling into a dedicated Prime Time dossier.
Fair-price and movement dashboard
Compare vig-adjusted market likelihood, model probability when available, break-even price and stored line movement. Movement is context—not automatic proof of sharp action.
Top researched player props
Daily prop research with price/freshness context. The full Player Intelligence explorer retains L20, book tree, market history and line-by-line detail.
Tests the official engine is built to consider
System-level capabilities reported by the live 4.2 model. Candidate cards above expose the evidence currently available for the same concepts without claiming a UI threshold overrides the engine.