One year, one chart

01 - Context

A season of lap times

Five track days at Thunderhill Raceway Park, West configuration, between 17 May 2025 and 12 April 2026. 115 laps across 16 sessions. The first four days carried 22 laps each and the fifth carried 27.

The fastest dry lap of the first day was 1:39.36. The fastest of the season was 1:28.39, set on lap 112 of the last day. That is 10.97 seconds. The steps between one day's best and the next were 2.24, 3.02, 3.10 and 2.61 seconds, which is the part I could not see while it was happening.

02 - Process

Recording first, the chart second

I was recording data before I knew what it was for. RaceChrono and RaceStudio logged every session from the first track day on, because logging is cheap and hindsight is not. There was no chart in mind yet, just laps going into a file.

The goal showed up once the data crossed a threshold. After multiple track days and over a hundred laps, the question stopped being about any single session and became the season: whether the time was actually coming down, and where it was coming from. Only the full set can answer that, and it set the design. Every lap in the order driven, one theoretical-best floor per day.

The published dashboard supplied the anchors: 115 laps, 16 sessions, a 1:28.39 best on lap 112, a 2:01.14 wet lap. The rebuild on this page models the remaining lap times to fit that distribution and says so in the source line under the plot. It stands until I export the original sessions.

03 - Design

Reading the plot

Each circle is one lap, plotted in the order driven. Seven of the 115 are wet and take the second series color. The rule under each track day is that day's theoretical best: 1:38.13, 1:35.89, 1:32.81, 1:29.70 and 1:26.93. No lap in the file reaches its own day's floor, and the gap from a day's best lap down to that floor runs from 1.23 seconds on the first two days to 1.46 on the fifth.

The All laps view covers 1:25 to 2:05, the range needed to hold the wet session. The Dry only view drops those seven laps and tightens the axis to 1:25 to 1:45, which gives the dry cluster more of the plot height. Nothing is clipped in either view.

Each theme has its own palette and the chart re-renders when the theme changes.

04 - Impact

The chart found the time I could not feel

Eleven seconds off a lap over five days, and none of it felt like progress from inside the car. Each day felt like the last one. The chart is where the shape shows up: the day-over-day steps are almost identical, 2.24, 3.02, 3.10, 2.61, which means the gains were coming from something repeatable and not from any one breakthrough session.

The theoretical-best floor under each day is the part I actually use. The gap from my best lap to that day's floor sits between 1.23 and 1.46 seconds and has barely moved across a year, so the time left is not in one corner I am blowing. It is spread thin across the whole lap. That changed what I practice.

This one also runs as my own regression test. It is the project where I check whether a chart still reads a year later, on a phone, in a theme I did not design it in first, and by keyboard alone. Every one of those caught something.