Last updated: 4 May 2026
Pitbrain is a one-person operation built on top of automated data pipelines and a small set of AI tools. This page lays out exactly how the games are scored, where the data comes from, and what the AI does (and doesn’t do).
Every standings table, calendar entry, and race result on the site is pulled from one of two free public APIs:
I write nothing about results until the official FIA timing is published. Game scores are computed against the official classification, not provisional.
Race and Quali predictions lock when the relevant session starts. After the session ends and FIA results are official, the scoring job runs automatically and pushes points into the leaderboard. If a position later changes (penalty, post-race investigation), the score is recomputed.
The fantasy game runs on a $100M budget with five drivers and a constructor. Prices update weekly based on form. After each race, points are calculated, posted, and visible on every team page. Boosters are one-shot, season-long.
Pit Wall answers F1 questions on the homepage. Most factual questions (standings, last race, career stats, head-to-head) are answered straight from the database in under 300ms with no LLM in the loop. Anything more open-ended falls back to Claude Haiku, grounded in the current season context.
Oracle is a separate pre-race prediction pipeline that runs about 30 minutes before lights out, generates a podium pick using Gemini Pro plus search, and scores it automatically once the race ends.
If something is wrong — a score, a stat, a name — tell me. The fix policy is on the corrections page.