São Paulo Grand Prix

The newest session shown here with lap timing is Singapore Grand Prix, Race (11 Oct).

Official results are shown where we have them. Anything built from lap timing, including predictions, may be less current than usual. This notice clears itself once lap timing returns to this page.

MET Norway forecast for each session's first hour, fetched at last regeneration (baked in, works offline). MET gives rain AMOUNT, not a rain chance.

Weather by session: the timing read

Circuit forecast (Interlagos) from MET Norway, one model (ECMWF HRES), fetched 2026-11-05 6:22 PM MST. A wet hour is 0.1 mm of rain or more (fixed in advance). MET Norway gives rain AMOUNT, not a rain chance, and no second model, so these reads carry no measure of their own uncertainty. Start times are official; end times are estimated from regulation lengths.

Race: 6-hour blocks only (too far ahead for a timing read). 12:00-18:00 UTC: 4.1 mm, rain; 18:00-00:00 UTC: 2.2 mm, light rain

How the forecast has moved, day by day

Weather data from MET Norway (the Norwegian Meteorological Institute), licensed under CC BY 4.0. Processed by this project into hourly session reads. Source: api.met.no. Licence: CC BY 4.0.

The next weekend has not started yet, so these are the results from Round 19, Mexico City Grand Prix, the most recent completed round. This section is replaced by the live weekend automatically as soon as the first session of the next round is collected.

Qualifying

Race

Official classification. Lap timing for this session is not available on this page, so there are no lap time or tyre columns.

Awaiting qualifying. The starting grid is published after qualifying and after any penalties are applied, so there is nothing to show yet. This section stays empty rather than showing a predicted order, because a prediction is not a grid.

No completed sessions yet this weekend.

Results and history data from jolpica-f1, licensed under CC BY-NC-SA 4.0. Processed and modified by us. Source: jolpi.ca. Licence: CC BY-NC-SA 4.0.

The circuit layout for Interlagos is not available at the moment. See the data delay notice at the top of this page.

2026 Round 13, Italian Grand Prix (Monza) qualifying (NOT this weekend). Reference lap GAS 81.786s. Across this session 91% of all time lost was mechanical, 9% was lost on straights, and 0% is unattributed : full throttle through faster curves, where grip and power limit the car at once. Mechanical plus energy is therefore not the whole story.

Precision. Gap is direct lap timing and keeps full precision. Mechanical, straight-line and unattributed are shown to the nearest 0.1s, because this split cannot tell two different cars apart by less than about 0.35s. The within-team contrast above keeps two decimals, because it compares two parts of the SAME car's lap and that limit does not apply to it.

Sections are classified by lateral load, not by curvature: a corner taken at 0.7g is not grip-limited when the car can pull 5g. The climbing-power term is removed using track altitude. See reports/grip_energy_split.md.

All computed from our data (11 races with lap/results on file). Small samples flagged.

Safety cars: 0 SC 25% · 1 43% · 2 31% · 3+ 1% from 7 races (small, UNVALIDATED)

Pit strategy | how this track rewards stopping early

Undercut / overcut: undercut-strong : the driver who pits FIRST gains +0.37 places on average (167 pit cycles, 4 races; F1-wide average +0.47)

Under a safety car: HIGH: a safety car rewrites the strategy here: first-stopper gain becomes -0.56 (a swing of -0.93) (280 caution cycles, 63% of all cycles here)

Descriptors, not predictions. Built from every season we hold, and read for insight. Neither has been tested as a model component. Caution cycles are excluded from the grade because a safety-car stop is a different thing (the field compresses and stopping becomes cheap).

Position change | does the grid hold, and do cars actually pass?

Read these three together, never just the first. Conversion says whether the finishing order differs from the grid. It does NOT say cars overtake: measured across 183 races, passing explains only about a third of it (R-squared 0.29), and the rest tracks the caution rate. Zandvoort is a worked example: the grid holds less than at 82% of circuits, yet it has more passing than only 35% and is more disrupted than 88%, so its position changes come from incidents rather than racing. Suzuka is the opposite: more passing than 77% of circuits, yet the grid holds less than at only 4%, because the passes simply restore the order pace already implies. Descriptive only, from 7 races here.

Circuit character | the five dimensions

Expert archetype: Barcelona-type (short undulating permanent)

DESCRIPTIVE, not a prediction, and not wired into anything. Circuits form a CONTINUUM on all five dimensions. The apparent groupings were single detached circuits at the edges of the range. So the continuous values and the position bar are the content; the band label is a reading convenience and is marked BORDERLINE when it sits close to a cut-off (70% of all labels do). Higher downforce means LOWER trap speed, so a low number = more wing. Grip is the weakest dimension: its two signals disagree on 17 of 31 circuits (here: signals DISAGREE).

Best current drivers here (avg finish)

Regulation first, because it can invalidate our data. Gossip and other series already stripped.

Weekend column source: awaiting practice.

Every completed round stays here after the page rolls over to the next weekend. Frozen predictions are kept in the database and are never overwritten; the linked files open in this browser.