Data Sources & Freshness

MLB research

Primary data: MLB game/schedule feeds and official lineup information, Baseball Savant / Statcast-derived research, historical model outputs and locally generated SlateSignals research files.

Refresh pattern: the daily workflow can be rerun as probable pitchers, lineups and weather become more reliable. Hitter research should be treated as more mature after confirmed lineups are available.

First inning: NRFI/YRFI is a separate model layer using team first-inning form, starter first-inning history, batting context and game environment.

NFL research

Primary data: verified nflverse historical game, roster, depth-chart, player-stat and play-by-play datasets, combined with SlateSignals Week-level modeling and manual market layers when applicable.

Refresh pattern: historical priors and active-player research are rebuilt before the weekly public export. Current-season data is incorporated as verified releases become available.

Market separation: player and game projections can exist without sportsbook lines. An edge is not shown as actionable unless the required market price is loaded.

Publication gate

Before a live deployment, SlateSignals checks required files, row counts, model readiness, current export dates and the presence of core MLB and NFL research datasets. A failed critical check stops deployment rather than replacing the live site with an incomplete build.

When a noncritical research layer is unavailable, the interface should show a clear waiting or partial-data state instead of presenting a blank table as if no players or games exist.

Sportsbook prices

Sportsbook lines can change quickly and may differ by state, account or time. SlateSignals keeps model probability separate from market price. Always verify the current line at the sportsbook before making a decision.

Historical outcomes and frozen pregame snapshots are intended for model evaluation, not as a guarantee of future performance.