Speedrunning—the practice of completing video games as quickly as possible—offers a unique window into how human performance evolves over time. Unlike traditional sports, it blends skill refinement, luck, and the discovery of new strategies or shortcuts. Despite the abundance of publicly available speedrunning data, there’s little research on forecasting record progression or modeling the underlying trends in improvement. Understanding these dynamics could provide insights into performance optimization, game design, and even the limits of human achievement in structured environments.

How the Idea Works

One way to study speedrunning records is by collecting data from platforms like speedrun.com and applying forecasting models to analyze improvement trends. This could involve:

• Log-transforming time data to account for exponential growth in player participation.
• Combining stochastic noise (from luck or variability in attempts) with extrapolated performance trends.
• Comparing record evolution across different games or genres to identify universal patterns or game-specific quirks.

The goal would be to create a framework that helps predict how records evolve, which could benefit speedrunners, game developers, and researchers studying human performance.

Potential Applications and Stakeholders

Speedrunners could use insights to refine strategies, while game developers might apply findings to design games that encourage or balance speedrunning. Researchers in human performance could generalize the results to other domains like sports or cognitive tasks. Gaming communities might also gain a deeper appreciation for the meta-level trends in their hobby.

Engaging stakeholders early—such as speedrunners for qualitative insights or platforms like speedrun.com for data access—could help bridge gaps between research and community interests.

Execution Strategies

A simpler version of this project could start by analyzing records for 2-3 popular games with long speedrunning histories, like Super Mario 64 or The Legend of Zelda: Ocarina of Time. Basic forecasting models could be applied, and preliminary findings shared in a blog post to gather feedback. For a more comprehensive approach, data collection could expand to 10-20 games across genres, with multiple modeling techniques tested and results published in gaming or interdisciplinary journals.

By focusing on the interplay between skill, luck, and iterative discovery, this idea could uncover new ways to understand and optimize performance in structured, rule-based environments.