Turing College Logo

Marathon Berlin Stats: Unofficial Race Insights

Marathon Berlin Stats: Unofficial Race Insights

About this project

An unofficial statistics site for the Berlin Mini-Marathon, the 4.2195 km school race held alongside the Berlin Marathon. It turns a raw finish list of more than 10,000 children into something kids, their families and teachers enjoy. Parents open a shared link on their phone and find their child in seconds. They come away with a story, not just a time: "faster than 73% of girls in the group", "beat the school median by a minute", or "fastest of the whole race".

Key features:

- Runner cards show the place in the group and in the whole field, how the child compares with classmates and the school, a "ghost race" against medians and winners, and fun facts.

- Earlier starts: children are linked across nine editions (2016-2026) using name, gender and school rules, so families can see progress year by year and race their earlier self on the interactive course map.

- Fair school rankings: team top-5, median times with minimum group sizes, turnout and finish rate, so small and large schools can be compared fairly. Free-text school names from older years are matched to the current list using the names and where the same kids ran the following year.

- "What if" tools show where a time would place in other groups or in earlier editions.

- Race replay on the real course map, simulated from finish times with a documented pace model and clearly labelled as a simulation.

- School recaps in the style of a sports commentator, written by Claude from a computed fact card. A script checks every number and name against that card before publishing.

The site is designed with privacy and kindness in mind:

- Surnames are reduced to an initial, and individual dots on charts are anonymous.

- Runners without a finish time are never suggested in search.

- Weak results are never highlighted.

It runs in English, Polish, and German, best UX on desktop, but works with mobile in mind. It is a static site with no backend and no third-party requests (self-hosted fonts for GDPR). It was built with Claude Code, from data analysis and data cleaning to the front end and deployment.

Created byTomasz Kornicki
Published atOctober 1, 2026
CourseBuilding with AI
Looking for
Freelance projectsNew jobCo-founderPotential users
OpenOpen LinkedIn