Runs (no laps)

When a recording is shown as runs

Laps need a loop: the car has to come back past the same place again and again. A drag pass, a hill climb, a rally stage or a drive from A to B never does, so no lap can be found in it. Such a recording is still full of data — speed, acceleration, cornering, where the car went — and it is shown as runs instead of laps.

A recording is shown as runs when it has GPS but no racing lap was found: either the lap detector refused it (not enough driving at racing speed to draw a track), or the laps it found were all pit, partial or invalid laps (the strip and the return road of a drag day make a loop whose "laps" are all stops). A recording with at least one racing lap is always a lap recording, exactly as before.

What a run is

The whole recording is put on one time grid of 10 points a second (the trace): the filtered GPS speed $v_{analysis}$ , the longitudinal acceleration $a_{long}$ and Cornering G $a_{lat,path}$ exactly as they are defined for laps (see Speed and Cornering G), the position, and the distance driven. Grid points inside a GPS gap longer than 1 s are left empty; no line is drawn across them.

The car is standing still when the receiver's speed $v_{valid}$ is at or below 0.5 m/s (1.1 mph). A run is a stretch of driving between standstills:

it ends where the car has stood still for at least 1 s; a shorter stop, and a GPS gap, are part of the run;
it starts at the last moment the car stood still before it moved, and ends at the first moment it stood still again (or where the recording or its GPS ends);
a stretch shorter than 100 m — a burnout, creeping in the pits or the staging lanes — is not a run.

A drag pass is therefore one run together with the shutdown and the return road, until the car next stops (at the time-slip booth, or back in the lanes). Its run time and distance say so; the pass itself is read from the drag-strip marks below, which are timed from the launch.

Distance

Distance is the receiver's speed integrated over time, not the sum of the steps between GPS positions: the speed is a velocity measurement (the receiver's Doppler solution), smooth and unbiased, while positions jitter by a metre or more from one sample to the next and would add distance even to a parked car.

$d = \sum_i \tfrac{1}{2}(v_{i-1} + v_i)(t_i - t_{i-1})$

between consecutive samples no more than 1 s apart. An interval where the speed is at or below the standstill speed at both ends counts as standing still (the receiver's speed never reads exactly zero). A GPS gap adds nothing.

Metrics of a run

metric	definition	shown
Run time	$t_{end} - t_{start}$	s
Run distance	$d(t_{end}) - d(t_{start})$	m
Top speed	highest $v_{analysis}$ in the run, away from the first and last half second of any stretch of GPS data (filter edges)	mph
Average speed	run distance / run time	mph
Peak acceleration	highest $a_{long}$ in the run, away from filter edges	g
Peak braking	lowest (most negative) $a_{long}$ in the run, away from filter edges	g
Peak cornering	largest $\lvert a_{lat,path} \rvert$ in the run, away from filter edges	g
Time into run, distance into run	at the cursor, since the run's start	s, m

A pass: drag-strip marks and 0–60 mph

When a run starts from a standstill the recording shows (not at the recording's start, and not after a GPS gap) and the car launches — it covers its first 60 ft within 4.0 s, 0.23 g on average from rest — the run is timed the way a drag strip times a pass, from the GPS samples themselves (10–18 a second, not the 10 Hz grid). A car rolling away from the pits or off a trailer does not launch, and is not timed.

The launch is the last standstill sample before the car first covers 60 ft: a roll to the line and a pause shorter than a second before the green are part of the run, not of the pass.
The distance is integrated from the launch, with no standstill floor, up to the first GPS gap. The speed is taken as linear between two samples, so the distance is quadratic between them, and the moment a distance is reached is solved exactly inside the interval that crosses it. The moment the car left, between the standstill sample and the next, is found by running the first two moving samples' speed line back to zero.
The clock starts when the car has moved 0.3 m — the rollout of a staging beam, about a foot — so a car creeping forward before it launches does not start it.
The pass ends at the first lift: once the speed has fallen 4.5 m/s (10 mph) below the best so far, no further mark and no 0–60 is timed. A drive that eases off before the eighth is not an eighth-mile pass; a pass pedalled harder than that ends where the driver lifted.
Elapsed time to a mark: $ET(D) = t(d{=}D) - t(d{=}0.3\,\mathrm{m})$ for the 60 ft, 330 ft, 1/8 mile (660 ft), 1000 ft and 1/4 mile (1320 ft) marks, measured from where the car stood.
Trap speed at the 1/8 and 1/4 mile: the average over the last 66 ft before the mark, as a strip's speed trap measures it, $v_{trap} = 66\,\mathrm{ft} / (t(D) - t(D - 66\,\mathrm{ft}))$ .
0–60 mph: from the start of the clock to the first moment $v_{valid}$ reaches 60 mph (26.8224 m/s), linear between samples.

A mark the run did not reach, reached only after a GPS gap or after the lift, is not shown.

A drag pass or a drive. A timed run is a drag pass when the pass was the fastest thing it did: once the car lifted, it never came back within 10 mph of the speed it lifted at — the shutdown and the return road of a drag strip. A hill climb or a drive from A to B that launched hard goes on as fast after its first corner: it is a drive, its run time and distance are its headline, and its times to 60 ft … 1/4 mile are shown as launch splits. Among the passes of a recording, the quickest time to each mark and the highest trap speed are marked best; the best pass is the quickest to the farthest trap mark any pass reached (1/4 mile, else 1/8 mile).

Display conversions

mph = m/s × 3600 / 1609.344; g = m/s² / 9.80665 (braking is negative). The marks keep their drag-strip names; distances are shown in metres. Nothing else is recomputed in the browser: the values at the cursor are the trace's own grid values.

Limitations

These are GPS estimates, not a timing system's. The receiver's speed lags true motion by 0.15–0.40 s; a constant lag moves the launch and the mark alike and cancels in an elapsed time, but the receiver's own smoothing does not, and a speed error of 0.1 m/s moves the 60 ft time by about 0.01 s and the 1/4 mile by about 0.02 s. Distances are from where the car stood, not from a surveyed starting line, and there is no reaction time: the clock starts when the car moves, not at the green light.
The 1 Hz speed filter rounds off a sharp peak, such as a drag car lifting at the finish line: for a pass, the trap speed is the better number than the top speed. Likewise the peak acceleration is the sustained pull, not the first instant of the launch.
GPS validation marks samples where the speed changes by more than about 1.2 g for longer than half a second as lower-confidence (they are still used), and rejects a step that reverses within a few samples as a spike.
A run is cut where the GPS is lost at its start or end; a gap in the middle is part of the run but adds no distance, and the marks after it are not measured (the run says how many gaps it had).

Verified

Synthetic drag passes at a known constant acceleration $a$ (so $ET(D) = \sqrt{2D/a} - \sqrt{2 \cdot 0.3/a}$): every mark within 1 ms at 10 samples a second, within 5 ms at 18, and within 30 ms with 0.05 m/s of receiver speed noise; trap speeds within 0.05 m/s; 0–60 mph within 1 ms.
A roll to the line and a 0.8 s pause before the launch (timed from the pause), a roll-away from the pits (not timed), and a lift before the eighth mile (the marks after it not timed).
A synthetic 5 km point-to-point drive with corners: one run, distance within 0.1 m of the truth.
Several drag passes in one recording (a loop the lap detector reads as pit laps): shown as runs, one per pass, each timed from its own launch.
A recording whose car never drove a run, and one with a single valid GPS sample: nothing is shown as runs.
Tests: run segmentation (short stops, burnouts, GPS gaps, a recording that starts while moving), the exact crossing solution, and the API and share-link documents.