Don't trust the win rate — check it yourself.
Most prediction sites show you a win rate you have to take on faith. The number lives in a database the operator controls, and nothing stops a losing pick from quietly disappearing or a probability from being nudged after the result is known.
Linerup works differently. Every day's full slate — every pick, every pass, the model's win probability and edge — is written to a public, append-only GitHub repository before first pitch. GitHub independently timestamps each commit, and the history cannot be silently rewritten. Anyone can confirm a prediction existed before the game started, and that no result was changed or deleted afterward.
The record lives here:
github.com/nickragone1-byte/linerup-recordTwo folders, mlb/ and nba/, hold one snapshot file per day — named snapshot-YYYY-MM-DD.json. Each file is the exact slate that was live on the site that day.
Open the record, go into mlb/ or nba/, and open any day's snapshot-YYYY-MM-DD.json.
Click Historyon that file. GitHub shows the exact UTC time it was committed — compare it to that day's first game and you'll see the commit lands before the games do.
Read the picks in the file and compare them to how the games actually finished. The prediction was frozen and timestamped before any of it played out.
Spot-check a losing day. Losses stay in the record exactly where they were committed — nothing is removed to flatter the win rate.
The very first commit in the repository is a one-time backfill, made on June 13, 2026. It imports the snapshots from before this public record existed, so those files carry the import date — not a true pre-game timestamp. Treat them as historical context, not proof.
Every commit after that backfill is the real thing: written automatically, before games, every day. Pre-game proof begins with the June 14, 2026 slate. We're telling you this because the entire point is honesty — quietly passing off the backfill as pre-game proof would defeat it.
A tracked record and a marketing number look identical until something goes wrong. The difference is whether the prediction was locked in, in public, before the outcome was known. Ours is — and you don't have to believe that, because you can verify it.