Racing on GLP-1: My Marathon & HYROX Results With Data

Thomas Prommer Hybrid Athlete & Technology Executive Follow on Strava

Published: March 27, 2026

Updated: April 9, 2026

Race Data Overview

This page collects my marathon results as baseline data — the races that made the weight-to-performance correlation impossible to ignore, and ultimately led me to start a GLP-1 protocol (tirzepatide/Mounjaro). Every finish time comes from official timing. Weight is race-morning scale weight.

I started GLP-1 in February 2026. None of the marathons below were run during the protocol — they represent the "before" picture. Upcoming races (IM 70.3 Da Nang, Challenge Roth) will provide the "during" data, and I will publish full results from each one.

My goal is transparency: I present both the numbers that look impressive and the ones that do not. Weight loss helps running economy, but race day is influenced by dozens of variables. I try to tease apart what changed because of weight, what changed because of fitness, and what was just weather and course conditions.

Baseline Marathon Data

All four marathons below are pre-GLP-1. I include body weight on race morning and conditions to provide context for each finish time. The pattern that emerges — faster at lighter weight — is what drove the decision to try GLP-1.

The standout comparison: Metropolis at 92 kg produced a 3:23 finish. Shanghai and Valencia at 95 kg produced 3:49 and 3:51 finishes — roughly 26-28 minutes slower. Berlin is the outlier: same 92 kg weight as Metropolis but a 3:55 finish, because conditions were hot and I paid for it in the second half.

This is observational data, not a controlled experiment. Fitness, taper quality, pacing strategy, and course profile all varied between these races. But weight was the biggest controllable variable, and the correlation was hard to ignore. When I found myself back at 94.5 kg in January 2026 despite consistent training, GLP-1 became a serious option.

HYROX Results

HYROX is a particularly interesting test case because performance depends on both running speed and functional station strength. Weight loss should help running splits but could compromise strength-dependent stations like sled push, sled pull, and wall balls.

I have not yet raced HYROX during the GLP-1 protocol. HYROX race data will be added as events are completed. The comparison I am most interested in: whether running split improvements from lower body weight offset any strength station losses from reduced mass.

Split Analysis

Raw finish times only tell part of the story. I do not have detailed split-by-split data with heart rate breakdowns for these races, but the overall weight-to-pace correlation across the four marathons tells a clear directional story.

Weight-to-Pace Correlation

The most striking pattern in the baseline data:

• At 92 kg (Metropolis): 3:23:41 finish, 4:49/km average pace
• At 95 kg (Valencia): 3:51:01 finish, 5:29/km average pace
• Difference: ~40 seconds/km slower at 3 kg heavier
• Over the marathon distance: ~28 minutes from a 3 kg weight difference

Berlin complicates the picture: I raced at 92 kg but finished in 3:55:07 due to heat. This is a useful reminder that weight is not the only variable. Conditions, taper quality, pacing decisions, and cumulative fatigue all play a role. The Metropolis-to-Valencia comparison is the cleanest weight signal in my data, but it is still not a controlled experiment.

Pace at Heart Rate

I do not run with a chest strap heart rate monitor or running power meter that would allow clean pace-at-HR comparisons across races. The Garmin wrist-based optical heart rate is too noisy for this kind of analysis. The marathon data gives directional evidence of the weight effect, but I cannot isolate it from aerobic fitness changes with the data I have.

For future races during the GLP-1 protocol, I plan to wear a chest strap to capture cleaner heart rate data. This will allow proper pace-at-HR comparisons between pre- and post-GLP-1 race efforts.

Power-to-Weight Progression

While I do not have running power data, my cycling power-to-weight ratio provides the clearest objective measure of how the GLP-1 protocol is affecting performance output relative to body mass. FTP estimates come from Intervals.icu based on structured cycling workouts.

The FTP gains are from structured cycling training (TrainingPeaks plan), not from the GLP-1 medication itself. But the W/kg improvement is amplified by simultaneous weight loss — I am producing more watts while carrying fewer kilograms. That is the compound benefit of combining structured training with body composition management.

Note: weight fluctuates daily. These are measurement-day values, not rolling averages. The current weight of 92.5 kg is higher than the 91.0 kg low point, which reflects normal daily variation and glycogen loading, not protocol failure.

For the full mathematical framework behind these calculations, see my race weight and power-to-weight deep dive.

Upcoming Races

The real test of this protocol comes with the next two races — the first endurance events at my new GLP-1 weight:

• IM 70.3 Da Nang -- May 10, 2026. First triathlon at the new weight. I will publish bike power data, run splits, and overall pacing. This is the first race where W/kg improvements should translate directly to measurable performance on the bike leg.
• Challenge Roth -- July 6, 2026. Full-distance Ironman. The big target. I expect comprehensive data: bike power over 180 km, run pacing over the marathon distance, and nutrition execution across 9+ hours of racing.

This page will be updated with full Garmin data, split analysis, and power numbers after each race.

Key Takeaways

After reviewing the baseline data, here is what stands out:

1. Weight is the biggest controllable variable. Across four marathons, the clearest predictor of my finish time was race-morning body weight. At 92 kg I ran 3:23. At 95 kg I ran 3:49-3:51. That is roughly 9 seconds per km per kilogram — directionally consistent with published estimates of 3-4 sec/km/kg for trained runners, though my numbers are higher, likely because conditions and fitness also varied.
2. Cycling W/kg tells a cleaner story. The bike power meter strips out pacing decisions, weather, and course profile. Going from 2.76 to 3.04 W/kg (a 10% improvement) in five weeks is significant. The FTP increase came from training; the W/kg amplification came from doing that training at a lower body weight.
3. This is not a controlled experiment. I changed multiple variables simultaneously: body weight, training structure, and training volume (triathlon preparation). Clean causal attribution is not possible. I am presenting correlations and being explicit about confounding factors.
4. The real data starts now. Baseline numbers establish the pattern. The upcoming races at new body weight will show whether the power-to-weight improvements translate to finish line results.

For the full protocol and journey context, return to the GLP-1 pillar page. For how I adapted fueling on race day, see my race day fueling guide.