RukiF1’s Formula 1 predictions are on pause for the rest of 2026. In 2027 we’re bringing them back, but with a twist: two contestants, one scoreboard.
The 2027 plan: The Human vs The Model
Every race weekend from the 2027 opener we’ll run two prediction pipelines side by side. One human editor who watches every session and picks by feel. One learning model that reads spreadsheets and has no feelings. Same scoring. Same table. Same public embarrassment. May the least-wrong contestant win.
The Human
Watches Friday practice. Watches Saturday qualifying and the sprint if there is one. Reads the news of the weekend (grid penalties, upgrades, weather, driver mood). Makes a top-10 call plus a Winner, a Big Surprise and a Big Flop. Has feelings. Has favourites. Occasionally has hunches that turn out to be nonsense. Will explain the loss in vivid detail.
The Model
Reads only what it’s given: recent-race form, qualifying pace at similar circuit archetypes, tyre-degradation history, historical head-to-head on this circuit, penalty status, season-long constructor pace trend. No opinions. No news scrape. No gut feel. Doesn’t know who Lewis Hamilton is. Only that driver_id 44 tends to score points. Loses gracefully because it can’t speak.
How we score them
Same rules for both. No handicap. Same scorecard, published Monday morning after each race.
| What happens | Points |
|---|---|
| Called an exact top-10 finishing position | +1 per hit |
| Nailed the Winner card | 0 (just for fun) |
| Called the Big Surprise correctly | 0 (still for fun) |
| Called the Big Flop correctly | 0 (bragging rights only) |
| Missed a top-10 position by one place | 0 (close doesn’t count) |
| Maximum score per race | 10 pts |
Ten top-10 slots, one point each, ten points on offer. The Winner, Surprise and Flop cards stay because readers like them, but they don’t score. Reality has too much random in it to score anything below tenth.
What each contestant does when they lose
| Category | The Human | The Model |
|---|---|---|
| Inputs | Every session, every rumour, weather app, gut | CSV files |
| Confidence | Unshakeable until proven otherwise | Precise to four decimal places, always wrong by three |
| Favourites | Yes and will deny it | None (a small tragedy) |
| Post-race excuse | “The safety car ruined everything” | Silence |
| Response to being beaten | A thousand words about tyre choices | A number that goes up |
| Sleep quality | Suffers | N/A |
| Advantage | Actual eyes | No emotions |
| Disadvantage | Actual emotions | No eyes |
Who wins the season decides who runs the show
Add up the points across every race weekend of 2027. Highest total at the end of the season is the primary pipeline for 2028. If The Model beats The Human, The Human gracefully accepts the outcome and shifts to writing an increasingly bitter Monday scorecard column. If The Human beats The Model, The Model does not mind either way, because The Model does not mind anything.
What happens to 2026 predictions
The prediction articles we ran in the middle of the 2026 season are no longer maintained. If you landed here from an old link, that’s why. The rest of the 2026 season is being used as the training window: we’ll keep generating internal predictions each race and score them privately, so The Model has some idea what a race looks like before it competes for real in 2027.
What we’re publishing in the meantime
Everything else keeps running. The weekly news roundup lands every Saturday. Race reviews go up Monday after each Grand Prix on the news page. Standings refresh Monday morning after every race on the standings page. Evergreen features across the season. Driver, team and circuit reference pages that get better every week.
When it all begins
The first live head-to-head prediction goes up ahead of the 2027 opener. If you want to see The Human and The Model publicly disagree about who wins the Australian Grand Prix, come back then. If The Model wins, we’ll be quiet about it. If The Human wins, we absolutely will not.