# Longitudinal Analysis and Predictive Modeling of Sprint Performance (1976–2035): Trends, Seasonality, and Prediction Approaches

## TL;DR

This study analyzes how sprint performance has changed over time and predicts future improvements using statistical models.

## Contribution

The study introduces a comparative analysis of ARIMA and regression models for predicting sprint performance trends.

## Key findings

- Sprint performance improvements have slowed since 2000, suggesting physiological limits may be approaching.
- ARIMA models predicted marginal improvements by 2035, with men's 100-m best time projected at 9.63 s.
- Regression models overestimated future gains compared to ARIMA, especially in the 400-m event.

## Abstract

This study presents a longitudinal analysis and predictive modeling of elite sprint performance trends from 1976 to 2035, based on a database of over 2,500 results from top 10 male and female finishers in the 100-m, 200-m, and 400-m events. Using regression analysis and time series models, including ARIMA and SARIMA, the study evaluated historical trajectories and predictions, accounting for seasonal effects related to Olympic-year cycles. Results indicated a significant long-term improvement in sprint performances, with the most rapid gains occurring before the year 2000. However, the rate of progress slowed, particularly in the 100-m and 400-m events, suggesting physiological limits may be approaching. ARIMA models predicted marginal improvements by 2035, with projected best times of approximately 10.67 s for women and 9.63 s for men in the 100-m event. Regression models, despite showing strong fits (R2 > 0.85), tended to overestimate future performance gains compared to ARIMA, particularly in the speed-endurance-dominated 400-m sprint. Comparative model assessments demonstrated that ARIMA provided superior predictive accuracy, better capturing historical variability and Olympic-cycle peaks. Practical implications suggested that future sprint performance gains would depend more on advancements in biomechanics, individualized training optimization, and sports technology, rather than on natural physiological improvements alone. This study highlights the necessity for integrating machine learning-based forecasting, biomechanical modeling, and strategic periodization to maximize sprinting potential in the coming decades.