tif1 serves timing and telemetry from 2018 onward from the TracingInsights data repositories. Results, lap times, pit stops and standings reach back to 1950 through the Ergast tables. Those come from the TracingInsights Stats repository, with the Jolpica API as fallback. This guide shows what that adds to a session and how to reach the tables directly.
Official classifications in a session
From 2018 onward every session carries the official Ergast classification next to the timing data. Readingsession.results or session.drivers fetches one small file for the round in parallel with the driver list.
- Race and Sprint:
Position,ClassifiedPosition,GridPosition,Status,Points,Time(the true total race time),Laps, and the extrasTimeGap,FastestLapRank,FastestLapNumber,FastestLapTime,FastestLapAvgSpeed. - Qualifying:
Position,Q1,Q2,Q3. - Every session:
DriverId,TeamId,PermanentNumber,DriverNationality,TeamNationality,DateOfBirth,DriverUrl,TeamUrl, andBroadcastName.
Position and GridPosition, as fastf1 does. session.drivers, session.drivers_df and session.results follow the official order, which includes post-race penalties.
Pit stops and standings
lib: pandas by default, polars with lib="polars". session.results follows it too: a PolarsSessionResults on a polars session.
Seasons before 2018
Schedules and sessions reach back to 1950. The schedule of those years lists all five sessions, as fastf1’s Ergast backend does, but onlyQualifying, Sprint and Race can be loaded. get_session(1995, "Monaco", "Practice 1") raises DataNotFoundError because Ergast carries no practice data. Results come from the classification, laps from the Ergast lap timings (1996 onward), pit stops from 2011 onward. Each lap that is faster than every earlier lap of the driver has IsPersonalBest set, so laps.pick_fastest() works as in later seasons. Telemetry, weather and race control messages do not exist for those years and raise DataNotFoundError. get_circuit_info() reads the earliest MultiViewer layout of the circuit, and returns empty frames for a circuit that MultiViewer does not know.
Before the 1960s a car was sometimes shared by two drivers, who then appear as two rows of the classification under one number. session.results keeps the plain number as the label of the first driver. The other drivers of that car are labelled number/Abbreviation, for example "10/shawe-taylor", so every label names one row. The DriverNumber column keeps the car number for both drivers. session.drivers and the keys of session.car_data list the same labels. session.get_driver() accepts a label, a number or an abbreviation: in the 1951 French Grand Prix, get_driver("8") returns Fangio and get_driver("8/fagioli") returns Fagioli, the second driver of car 8.
Time and LapStartTime are the cumulative race time (Ergast has no session clock). Stint follows the pit-stop laps, and channels Ergast does not know (sectors, tyres, speed traps) are empty. Drivers without an Ergast three-letter code (most drivers before the 1990s) keep their Ergast id as Abbreviation and as the Driver value of the lap rows.
The schedule of a season before 2018 is built from the Ergast race table: practice sessions are listed with estimated dates, as fastf1’s backend="ergast" does. Ask for that schedule for any season:
CircuitId, CircuitName, CircuitLatitude, CircuitLongitude, EventUrl and CircuitUrl after fastf1’s columns.