SlateSignals separates source data, model estimates, uncertainty and decision thresholds so the research can be evaluated instead of treated as a black box.
The game model uses opponent-adjusted offensive and defensive efficiency, expected drives, recent form, scoring rates and 10,000 simulations to estimate team scoring, totals and win probability. Reliability measures evidence depth rather than pick strength.
Player projections are built from stabilized opportunity volume multiplied by regressed efficiency, with pace, game script, current role and category-specific opponent allowance. The raw projection is treated as a Model Fair Line, not a promise that the player will finish on that exact number.
SlateSignals converts the Model Fair Line and player-specific empirical volatility into simple at-least thresholds such as 20+, 30+, 40+ receiving yards or 4+, 5+, 6+ receptions. Yardage ladders move in 10-yard steps and reception ladders move one catch at a time.
Strong/Elite projection rows automatically flag the highest milestone still carrying at least a 70% model probability so the same signal can be archived for accountability.
The TD model remains separate from yardage/reception projections and uses scoring history, red-zone opportunity, inside-10 usage, team scoring environment and opponent positional TD allowance.
Pregame snapshots are archived before results are known. For player markets, O/U correctness is the primary market-facing grade whenever an archived market line exists. Exact-stat mean absolute error is retained as a secondary diagnostic for projection drift.
Historical rates describe what happened. Model Fair Lines estimate the center of a distribution. O/U probability estimates the chance of beating a supplied threshold. Reliability describes the evidence behind the estimate. These are intentionally different concepts.
NBA V1 is an experimental minutes-based model using supplied roster, rotation, schedule and log inputs. Its probabilities are uncalibrated; the page remains waiting until data is loaded. NCAAF uses the existing College Football Bible scoring model and published reports. Refresh timestamps and data limitations are shown inside each sport.
