Variance collapse

The expert is only 0.5s faster on their best lap. But 4.0s better on their worst. That difference is dominance.

Worst laps determine position more than best laps. One bad corner costs multiple positions.

Best laps often include luck. Worst laps reveal gaps that luck can't cover.

Low variance means the skill is internalized and automatic. High variance means you're still consciously managing.

Lap times can plateau while variance shrinks. That's real improvement even if the number on screen doesn't change.

Standard deviation — measures typical deviation from the mean. Lower is more consistent. This is the primary metric I track.

IQR — interquartile range. Middle 50% of laps. Ignores outliers.

Range — best minus worst. Simple but sensitive to single outliers.

What to track: lap time variance across sessions, sector variance (which sector swings most?), session-to-session reproducibility.

1. Identify high-variance zones from sector or corner data

2. Diagnose the cause — usually inconsistent braking, sloppy turn-in, or reactive exit corrections

3. Create anchor points — physical or visual references that make inputs repeatable

4. Drill to stability — repeated runs hitting anchors. Not fast, correct. Speed comes from repeatability

5. Test under load — stable in isolation, then full laps, then stints, then traffic

Focus on consistency first. Build a reliable baseline (low variance, not necessarily fast). Push speed in one area. Stabilize at the new level. Push the next area.

Pushing speed before variance is controlled means building on an unstable foundation. Gains won't stick.