Performance prediction equation for the Valencia Marathon based on time and pacing in the half marathon

Area of Science:

• Sports Science
• Running Performance Analysis

Background:

• Marathon running performance is influenced by pacing strategies.
• Existing models for marathon time prediction lack comprehensive analysis of pacing variables.

Purpose of the Study:

• To develop a linear regression model for predicting marathon race time using half marathon data.
• To assess the impact of pacing range (variability in speed during the half marathon) on prediction accuracy.
• To compare the developed model's accuracy against the established Daniels' VDOT system.

Main Methods:

• Utilized data from 8,261 runners participating in both the Valencia Half Marathon and Marathon in 2022 and 2023.
• Developed three linear regression models, progressively adding variables: half marathon time, sex, age category, and pacing range.
• Evaluated model accuracy using explained variance and mean absolute error, comparing the best model with the VDOT system.

Main Results:

• The final linear regression model, including half marathon time, sex, and age category, achieved 85% explained variance and a 5.9% mean absolute error.
• Incorporating the pacing range variable did not significantly enhance the prediction model's accuracy.
• The developed model demonstrated similar overall accuracy to the VDOT system, though performance varied across different runner ability levels.

Conclusions:

• Half marathon time, sex, and age category are key predictors of marathon performance.
• Pacing range is not a significant factor in improving marathon time prediction accuracy with this model.
• The findings provide valuable insights for coaches and runners to optimize training and race strategies based on performance predictions.