Pick the question first
A stint comparison answers one of a few questions, and each needs different laps. Mixing them is the most common reason a delta misleads.
| Question | Compare | Why |
|---|---|---|
| Is run B slower overall? | Run medians of clean laps | One lap can hide traffic or a mistake. |
| Where is the time lost? | One representative lap per run | A trace needs a real lap, not an average. |
| Did the setup change help? | Clean-lap medians at similar fuel and tyre state | Fuel burn and tyre wear move lap time on their own. |
| Was a lap affected by traffic? | Your note, then your decision | Traffic is a note, never an automatic exclusion. |
Match the conditions before the laps
Two stints are comparable only when the car and the track are in a similar state. Check these before reading any number.
| Condition | Why it matters | If it differs |
|---|---|---|
| Fuel load | More fuel costs lap time and shifts balance. | Compare laps at similar fuel, or state the difference next to the delta. |
| Tyre state | New, scrubbed and worn tyres give different grip. | Compare laps of similar tyre age, or split the stint. |
| Track state | Rubber, temperature and wind change grip through a session. | Prefer stints run close together in time. |
| Traffic and flags | A lap behind another car or under yellow is not the car's pace. | Exclude the lap and write down why. |
| Driver | Two drivers take different lines and inputs. | Compare the same driver unless the driver is the question. |
Exclude laps on purpose, and say why
Remove out-laps, in-laps and laps with traffic, flags or an off. Decide by judgment, not by a lap-time threshold: a slow lap is not always unrepresentative, and a fast lap can hide a tow.
Write the reason next to every exclusion. The next person to read the comparison, often you a week later, needs to know why the lap is missing.
Use medians for pace
With a handful of clean laps per stint, the median is the safer summary. One mistake drags a mean; it barely moves a median. Laps of 104.2, 104.3, 104.4 and 106.9 s have a mean of 104.95 s and a median of 104.35 s.
State how many laps each median covers. A median of two laps is a weak summary, and the reader should see that.
Use one representative lap for traces
Averaging traces smears braking points and apexes across laps, so the average describes a lap nobody drove. Pick one real lap per run that sits close to the run median and is clean through the section you care about, then compare lap against lap.
Keep the representative laps separate from the medians. Excluding a lap changes the median; it should not quietly change the trace you are reading.
Worked example: the synthetic Jerez session
The synthetic session in Racing has two runs of three laps each on a fictional GT4 car. Run B moved the front roll bar from position 3 to 4 and started with 32 L of fuel against 35 L for run A. B03 ran into traffic.
- Exclude B03 and record traffic as the reason. Run B's median now covers two laps; say so next to the number.
- Note the 3 L fuel difference. The lighter car should be slightly quicker, so fuel does not explain a slower run B.
- Read the median delta for pace, then compare A02 and B02 through sector 2 for where the time went.
- Hot tyre pressures are missing for both runs. Record them as missing rather than assume they matched.
Checklist before you read a delta
- The question is written down: pace, location or a setup effect.
- Fuel, tyre state and track state are close, or the difference is stated.
- Every excluded lap has a written reason.
- Each median states how many laps it covers.
- Traces compare one real lap per run, not an average.
- Missing measurements are listed as missing.