From the Athens Marathon to the Athens Half Marathon: race-time correspondence and course pacing in linked entrants, 2024/25 and 2025/26 race sequences

Background. Many Athens Marathon finishers enter the Athens Half Marathon four months later. We examined linked race times, Riegel conversion error and observed Half Marathon pacing, with attention to selection and timing-point uncertainty. Methods. Public records from the 2024/25 and 2025/26 sequences yielded 1,585 execution- and level-screened pairs, 2,491 identity-only pairs and 4,300 balanced pacing cards. Association models adjusted for edition, with runner-group clustering. The plan was locked after initial inspection, before final fits. Tools used separate fitting, calibration and holdout groups, one final acceptance evaluation and later descriptive rechecks. Results. Riegel 1.06 predicted Half times slower than observed by a mean 12.4 minutes (95% CI 11.9 to 12.8) in the main cohort and 14.6 (14.0 to 15.2) in the identity-only cohort; the miss grew with Marathon time. At the same first Marathon half, a 10 percentage-point greater second-half slowdown was associated with a 2.0% (1.6 to 2.4) slower subsequent Half. The 10 to 15 km section was faster and the 15 to 20 km section slower than mean pace across editions, finish bands and both distance axes. Holdout Target mean absolute error was 5.5 minutes (Riegel 12.3). Its original nominal 80% range covered 91.1%, failing the 75 to 85% criterion. A later descriptive central P25–P75 recheck covered 46.8%. Splits error was 50.7 seconds, versus 64.0 for constant pace and 53.9 for grade-adjusted pace. Conclusions. These selected linked performances support course-specific descriptive references. Outcome-dependent selection and cohort composition limit generalisation. The Target typical time improved on Riegel in the internal holdout, but its original range failed the coverage criterion; the later central range is not independently validated. The stable pacing pattern does not establish an optimal strategy. Neither tool estimates personal success probability or guarantees future-edition accuracy.

Authors

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-11-14
DOI
https://doi.org/10.5281/zenodo.24223318
Primary Topic
Sports Performance and Training
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

From the Athens Marathon to the Athens Half Marathon: race-time correspondence and course pacing in linked entrants, 2024/25 and 2025/26 race sequences

Georgios Svarnas
Zenodo (CERN European Organization for Nuclear Research)
Sports Performance and Training
preprint

From the Athens Marathon to the Athens Half Marathon: race-time correspondence and course pacing in linked entrants, 2024/25 and 2025/26 race sequences

Georgios Svarnas
preprint en

Abstract

Background. Many Athens Marathon finishers enter the Athens Half Marathon four months later. We examined linked race times, Riegel conversion error and observed Half Marathon pacing, with attention to selection and timing-point uncertainty. Methods. Public records from the 2024/25 and 2025/26 sequences yielded 1,585 execution- and level-screened pairs, 2,491 identity-only pairs and 4,300 balanced pacing cards. Association models adjusted for edition, with runner-group clustering. The plan was locked after initial inspection, before final fits. Tools used separate fitting, calibration and holdout groups, one final acceptance evaluation and later descriptive rechecks. Results. Riegel 1.06 predicted Half times slower than observed by a mean 12.4 minutes (95% CI 11.9 to 12.8) in the main cohort and 14.6 (14.0 to 15.2) in the identity-only cohort; the miss grew with Marathon time. At the same first Marathon half, a 10 percentage-point greater second-half slowdown was associated with a 2.0% (1.6 to 2.4) slower subsequent Half. The 10 to 15 km section was faster and the 15 to 20 km section slower than mean pace across editions, finish bands and both distance axes. Holdout Target mean absolute error was 5.5 minutes (Riegel 12.3). Its original nominal 80% range covered 91.1%, failing the 75 to 85% criterion. A later descriptive central P25–P75 recheck covered 46.8%. Splits error was 50.7 seconds, versus 64.0 for constant pace and 53.9 for grade-adjusted pace. Conclusions. These selected linked performances support course-specific descriptive references. Outcome-dependent selection and cohort composition limit generalisation. The Target typical time improved on Riegel in the internal holdout, but its original range failed the coverage criterion; the later central range is not independently validated. The stable pacing pattern does not establish an optimal strategy. Neither tool estimates personal success probability or guarantees future-edition accuracy.

Zenodo (CERN European Organization for Nuclear Research)
Sports Performance and Training
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon