Wharton Data Science: From Football Stats to Defensible Predictions

Editorial overview

The Wharton High School Data Science Competition is organised by the Wharton Sports Analytics and Business Initiative with support from Wharton Global Youth. Wharton uses sports analytics to give students experience in data analysis, prediction and teamwork. That combination explains the opportunity better than the university name alone: a team must turn a sporting question into an analytical argument that other people can inspect. Interest in football can supply the initial curiosity, but the worthwhile work lies in deciding what the numbers mean and how confidently a prediction can be defended.

The challenge is particularly appealing to a school group that wants a shared, bounded data project rather than a collection of unrelated coding exercises. Students who enjoy explaining a graph may contribute as much as those who enjoy writing a model. Our view is that the demanding part is the handover between those activities: a convincing result needs a sound method, readable evidence and a team that understands both. Families should expect sustained collaboration and revision, including some results that do not support the team's first idea. Choose it when students want to investigate and explain together; reconsider when the attraction is simply attaching a familiar university name to an activity.

Quick Facts

Review Evaluation

For current dates, eligibility and registration details, see the Wharton High School Data Science Competition 2027 competition page.

Review Evaluation

Rated Advanced. The challenge is choosing sound analytical methods, producing clear visualisations and explaining the reasoning behind results in concise slides and, for selected teams, live presentations.

A fictional league makes the method visible

Use simulated player statistics from a fictional professional American football league to rank players, predict outcomes, calculate performance statistics and produce data visualisations.

That is a useful distinction for students who already follow the sport. Familiarity with famous players cannot settle the problem. A team must work from the supplied information and decide which patterns deserve attention. The setting invites sporting questions, while keeping the evidence attached to a dataset everyone can examine.

A practical starting point is to read the data dictionary together and explain the unit of observation in ordinary language. Does a row describe a player, an appearance or a period of performance? What does each variable measure? Which comparisons would be misleading without accounting for opportunity or role? These are preparation suggestions, rather than claims about the unreleased data. They illustrate the habits that help a team avoid building an elaborate calculation on a misunderstood column.

Before choosing a complicated technique, write down the prediction being attempted and a simple comparison method. Keep a record of the assumptions. If a more elaborate model changes the result, ask which feature or decision caused the change. A method the team can interrogate is more useful than an impressive label that nobody can explain.

The slide limit changes the analysis

Three phases: initial football analysis, semifinal slide presentation, and live virtual presentations by five selected teams.

The slide deck is limited to seven slides using the provided template; selected teams use that same deck in their live presentation.

Develop explanation alongside computation. A chart should help someone understand an analytical choice, rather than decorate a result. Conversely, attractive graphics cannot repair an unsupported comparison. For this competition, the small slide allowance makes editing part of the analytical work: students must decide which evidence carries the argument and which details belong in their working notes.

One useful rehearsal is to give a teammate a chart without commentary. Ask what they think it shows, where the uncertainty sits and what decision they would make from it. If their interpretation differs from the author's, improve the labels or the chart before adding more explanation. This test is especially valuable when the person who wrote the code has become too familiar with its assumptions to notice missing context.

Keep the analytical record more detailed than the presentation. The audience needs a clear route through the reasoning, while the team needs enough working material to check calculations and answer questions. Those are different writing tasks. Trying to put every experiment on a slide usually makes both harder.

Share the reasoning across the team

Teams maintain 3–5 contributing students from the same high school and branch, with one school teacher/educator Advisor; students may join only one team. Home-school teams need prior organiser approval and documentation under the official exception.

Check the current age limits and the separate Student Team Leader minimum before choosing the team. The organiser distinguishes current pre-university students from school graduates; do not assume that enrolment alone settles eligibility.

The school Advisor registers the team and adds the Student Team Leader as a collaborator; students may not register a team themselves.

Those arrangements make school coordination part of deciding whether to participate. Establish the adult contact and the team's working routine before allocating technical tasks. For home-school circumstances, ask the organiser to confirm the permitted structure rather than assuming that any informal group is eligible.

Within the group, divide work without dividing understanding. One student might investigate variables, another test a method and another draft a visual explanation. Then exchange the work. Everyone should be able to describe why the team chose its approach and where it may fail. A polished presentation assembled from isolated contributions can conceal misunderstandings until someone asks a basic question.

We would also recommend a short decision log: what changed, why it changed and which evidence prompted the change. It supports a more honest account of the investigation than reconstructing a tidy story after the results are known.

Tools can assist, but students must explain

Generative AI is permitted, but teams remain responsible for mistakes and must explain the logic and reasoning behind their approach. Cite sources and acknowledge ownership of images and other media; submitted analysis must be the students’ own work.

That rule gives the team a clear practical test: can a student explain and check the output without treating the tool as an authority? Suggested code still needs inspection; a generated explanation still needs comparison with what the analysis actually did. Keep any tool-assisted work traceable enough that teammates can review it. When uncertain about a particular use, ask the competition staff before incorporating it.

For parents, the best support is organisational: help students protect meeting time and notice when the workload has become uneven. Avoid taking over interpretation or polishing the argument into an adult voice. The value of this task is the team's growing ability to connect a question, a method and an explanation.

Key Takeaways

Choose Wharton's data competition for a team that wants to investigate sport through evidence and can sustain a shared project. The main preparation is learning to question the data, compare methods and explain decisions together. Treat the slides as a test of understanding: if the reasoning cannot survive clear, concise explanation, the analysis probably needs another pass.

Sources checked

Checked on 2 November 2026. Consult the competition page and official sources for current arrangements.

• https://globalyouth.wharton.upenn.edu/competitions/data-science/
• https://wsb.wharton.upenn.edu/wharton-data-competition/
• https://wsb.wharton.upenn.edu/wharton-data-competition/rules/
• https://wsb.wharton.upenn.edu/wharton-data-competition/timeline-and-dates/
• https://wsb.wharton.upenn.edu/wharton-data-competition/deliverables-and-judging-criteria/
• https://wsb.wharton.upenn.edu/wharton-data-competition/faqs/