cortex-rubric: transparent matchup analysis for every game
cortex-rubric is the explanation layer for WhoWins.ai. It turns noisy pregame context into a consistent set of matchup factors so users can see why a game leans one way, where the signal is thin, and what uncertainty remains.
What cortex-rubric does
cortex-rubric gives every eligible matchup the same structured read. Instead of producing a freeform opinion, it scores a fixed set of factors, identifies top drivers, surfaces uncertainty flags, and creates a concise summary that can be displayed consistently across web and iOS.
Why it is separate from Hermes
The prediction percentage and the written analysis answer different questions. Hermes NBA 1.1 focuses on probability. cortex-rubric focuses on explanation. Keeping those layers separate lets WhoWins.ai show a model percentage while still explaining the game through market signal, team context, injuries, venue, schedule, and volatility.
The 12-factor checklist
The rubric covers market signal, team strength, recent form, injuries and availability, starting lineup or key-position edge, rest and schedule, home or venue edge, matchup fit, motivation context, coaching and tactical edge, weather or external conditions, and volatility or uncertainty. When evidence is missing, the factor stays neutral rather than being guessed.
How scores should be read
Each factor is scored from -2 to +2, where negative values favor the away side, positive values favor the home side, and zero means neutral, unsupported, or inconclusive. The goal is not to make every factor dramatic. The goal is to make the few meaningful drivers easy to find.
What users should take away
cortex-rubric is designed to make AI analysis more inspectable. If the data is strong, the summary can say so. If the data is missing or unstable, uncertainty remains visible. That makes the scorecard easier to compare against Hermes probabilities, market prices, and community predictions.
Keep reading