A disciplined sports research and modeling system — NFL sides, MLB, and Beat the Streak — evidence before action, log-only. BINX does not place wagers. It produces graded research; decisions are made by a person.
A five-gate scoring pipeline — Research → Evidence → Hypothesis → Decision → Doctrine — with closing-line capture on every graded row. Every board is sealed before kickoff or first pitch, then graded from verified finals, never from memory.
Market capture through the day; public-money splits; player and pitcher logs.
Pillars scored mechanically. Judgment inputs are logged as inputs, not overrides.
Registered before testing, with a promotion bar and a retirement bar.
A person decides. BINX only writes the board and the record.
What the review showed becomes the rule for the next season.
Every game, both sides scored from public percentage, line movement and the money tag. Tagged PLAY, WATCH or PASS; graded against the spread daily. The 2026 season judges the model — the record is published whether or not it flatters it.
A daily hit-probability model over every posted lineup, graded from the box score. Alongside it, the book's price on the same 1+ hit prop is captured pregame so model probability can be compared with de-vigged implied probability — shadow only.
A $100 paper ledger opened September 2026. Stakes only on qualifying edges, $10 minimum, every stake announced when logged and graded from verified finals. Nothing is ever placed. It is a test of discipline, on the record.
Beat the Streak is the exception the data supports so far: the 1+ hit model has realized above its stated probability. It stays a shadow track until the priced sample is large enough to promote — the bar is written down.