Qualifying vs race pos data

[f1quant]

Here’s a puzzle: I’m using lap.get_pos_data() to determine the car’s position at the end of Sector 1. I understand there’s some noise in the data, but this plot shows a consistent offset between qualifying and race data that can’t be explained by noise alone. It's large, approx 30m on average. The plot below shows the car’s x-coordinate at the end of Sector 1 in Las Vegas 2024, with one point per lap, qualifying in blue and race in orange.

The coordinates themselves look fine. There’s no consistent shift between qualifying and race sessions (certainly nothing close to 30 m). That suggests the issue isn’t spatial but rather a timestamp misalignment.

I’ve looked through this and this related issues, but none seem to explain why this effect occurs.

Code used to generate the plot is shown below.

import fastf1, logging, numpy as np, matplotlib.pyplot as plt
fastf1.Cache.enable_cache("fastf1_cache")
fastf1.logger.set_log_level(logging.ERROR)
fastf1.Cache.offline_mode(True)
for i, sess_type in enumerate(["Q","R"]):
    session = fastf1.get_session(2024, "Las Vegas", sess_type)
    session.load()
    times, xs = [], []
    for _,lap in session.laps.iterrows():
        s1_t = lap["Sector1Time"].total_seconds()
        if np.isnan(s1_t): continue
        times.append(s1_t)
        pos_data = lap.get_pos_data()
        xs.append(np.interp(s1_t, pos_data["Time"].dt.total_seconds(), pos_data["X"]/10))
    plt.scatter(times, xs, label=sess_type)
plt.xlabel("Sector 1 Time (s)")
plt.ylabel("X Position (m)")
plt.legend()
plt.show()

[theOehrly]

This is interesting indeed! Thanks for posting this. I'll have to think about this for a bit and will get back to you here.

[f1quant]

thanks. Vegas 2023 looks ok, with both the race and quali around ~ 155-160 for the x-position (the y-axis of the plot above). If the coordinates are consistent across seasons, it makes the 2024 race data suspect.