Kimi Antonelli had 42 lap times deleted in the 2025 season, almost one for every race he started. The next highest was 27.
In the same Mercedes, George Russell lost 22.
And the thing everyone assumes about this, that rookies are the ones running wide, is wrong. Take Antonelli out and the 2025 rookie class was cleaner than the established drivers.
Lap times deleted per race started for every driver in the 2025 season
What we counted
Every lap in Formula 1 carries a flag recording whether the time was deleted, and a reason. We pulled that flag for every qualifying session and every race of 2025.
Raw totals are unfair, so everything here is per race started. A driver who contested seventeen rounds cannot be compared directly against one who did twenty-four. Normalising changes the order: Franco Colapinto has 23 deletions in total, fewer than Gasly's 27, but a worse rate because he raced less.
The season
Antonelli — 1.83 per race (42 total)
Colapinto — 1.35 (23)
Gasly — 1.17 (27)
Hamilton — 1.12 (27)
Bearman — 1.04 (25)
Sainz — 1.00 (22)
Norris — 0.96 (23)
Russell — 0.92 (22)
Leclerc — 0.83 (20)
Verstappen — 0.65 (15)
Piastri — 0.57 (13)
Ocon — 0.42 (10)
Alonso — 0.42 (10)
Antonelli's 1.83 is not just top of the list, it is 36% clear of second place. Over a season that is a lap time lost at nearly every event.
At the bottom, Alonso and Ocon lost a lap time roughly every other race. Alonso being the cleanest driver on the grid, at 44 years old and in his twenty-second season, is the kind of result that makes intuitive sense and is rarely actually checked.
Same car, twice as many deleted laps
The car matters here. Some cars are more stable on kerbs, and a team's circuits are identical to their team-mate's. So the honest comparison is within a team.
Deleted lap times per race compared between team-mates in 2025
Every single pairing has a gap, and four of them are close to double or more. Identical machinery, identical circuits, and one driver over the white line twice as often. Whatever this is measuring, it is a property of the driver.
Norris against Piastri is worth noting, because it agrees with something we measured completely independently. Piastri loses lap times at 0.57 a race against Norris's 0.96. He was also the more consistent of the two on lap time, at 0.329 against 0.468. Two different measurements, from two different datasets, pointing the same way about the same pair of drivers.
Rookies are not the problem
The obvious story here is inexperience. It does not survive the data.
Deleted lap times per race for the 2025 rookie class against established drivers
The 2025 rookies averaged 0.99 deletions a race. The established drivers averaged 0.84. A gap, but a modest one.
Then remove Antonelli. The remaining rookies average 0.78, which is better than the established drivers.
The entire apparent rookie effect is one driver. Lawson at 0.57 and Bortoleto at 0.65 were cleaner than Hamilton, Norris, Sainz and Gasly. Bearman at 1.04 is high, but he still lost lap times less than half as often as Antonelli.
This matches what we found looking at lap time consistency, where the rookie penalty came out at 0.031 seconds and Bearman finished fourth on the entire grid. The "scrappy rookie" is a much weaker pattern than it is assumed to be.
What this is and is not measuring
A deleted lap time is not a mistake. As we found looking at which corners produce them, deletions cluster at corners where running wide is quicker. A driver who never has a lap deleted may simply be leaving time on the table.
That reframes the whole table. A high number could mean a driver is untidy, or it could mean they are probing the limit harder and occasionally being told no. Antonelli at 1.83 a race, in his first season, is at least as consistent with a young driver exploring where the edge is as with one who cannot find it.
It is also circuit-weighted. Every driver raced the same calendar, so this is fair between drivers, but a season with more Abu Dhabis and Monacos would produce more deletions for everyone.
And it counts qualifying and races together. Races produced 299 of the 408, qualifying 109.
What this changes about watching a race
Watch for the repeat offenders at the known corners. Antonelli at Abu Dhabi turn 1 is a different proposition from Alonso there.
Do not read a deletion as incompetence. At the corners where these happen, going wide is faster, and the driver is finding out whether it counts.
Team-mate gaps are the real signal. Same car, same circuits, and a factor of two between drivers is a genuine difference in how they use the track.
Method
Data: FastF1, every qualifying session and race of the 2025 season.
Deletions from the per-lap `Deleted` flag with the reason text attached. 408 deletions, 403 explicitly citing track limits.
Normalised by races started, taken from race classifications. Drivers with fewer than 10 starts are excluded, which removes Jack Doohan (7 deletions).
The rookie group is Antonelli, Bearman, Bortoleto, Hadjar and Lawson, the five full-season 2025 debutants who met the appearance threshold.
Team-mate comparisons use drivers assigned to their most frequent team, which is why the Red Bull and Racing Bulls pairings are the least clean.
This measures deletions, not errors. The distinction is set out in the body and matters for interpreting every number here.
Qualifying and race deletions are combined. The split is 109 and 299.
