HYROX's stall accelerates due to the "90 minute wall"

Following #1 and #2, this third article focuses on pacing. Public data shows late-race slowdown accelerating around the 90-minute mark, so we organize practical training priorities to break through that wall.

Main points of this article

  1. Run8 stall rate is+20.0% with Sub-70+27.6% with Sub-90+37.2% with Sub-100
  2. The difference between Sub-90 → Sub-100 (+9.6pt) is larger than the difference between Sub-60 → Sub-70 (+3.1pt), and the stall increases non-linearly.
  3. In the Quantile analysis, the coefficient of Run8 isQ10:2.71 → Q90:3.35(+23.6%)increased to
  4. In the Sub-100 zone, it is more effective in the short term to create a distribution that does not collapse in the final stages than to run fast.

table of contents

  1. Why does the “90 minute wall” separate records?
  2. Stall pattern that tends to occur around Sub-100
  3. Where should I start to grow in the shortest time?
  4. 4 week plan (90 minutes wall countermeasures)
  5. Race day running rules
  6. Common mistakes and how to fix them
  7. Three numbers to track with HYFIT
  8. FAQ

1. Why does the “90 minute wall” separate records?

HYROX cannot push through with just the momentum of the early stages. In the final stages of the race when fatigue piles up, when the run, event, and judgment start to fail at the same time, The stalls will chain together and your time will drop all at once.

Data shows that stalling is not linear. There is a tendency for the rate of decline in Run 8 to suddenly increase after around 90 minutes. This is the so-called "90 minute wall."

Decreased pacing by level of HYROX. Stall tends to accelerate around 90 minutes
Figure 1. Pacing decline is nonlinear. It tends to fall apart around 90 minutes.

What this diagram shows is not a "lack of guts" but a problem with allocation design. In other words, even with the same running power, it is possible to stay in front of the wall by simply changing the distribution.

2. Stall pattern that tends to occur around Sub-100

The head difference of Run 8 based on Run 1 is especially large for Sub-100. Furthermore, as shown in the figure below, it can be seen that the slower the layer, the greater the variation (width of stall swing).

Run pace transition curve by level. There is a large decline in the final stage in the Sub-100 band.
Figure 2. Change in pace by level. There is a large decline in the final stage in the Sub-100 band.
Distribution of run stall rates. The slower the layer, the wider the dispersion.
Figure 3. Variance of stall rate. The slower the layer, the lower the reproducibility.

The following three things are common among players aiming for Sub-100.

Here, it is more important to improve reproducibility in the final stages than absolute speed.

3. Where to start to grow in the shortest time?

In Quantile regression, the "place where it works" changes between the fast layer (Q10) and the slow layer (Q90). In particular, the Run8 coefficient increased from 2.71 to 3.35 (+23.6%), indicating that the slower the layer, the greater the influence of run collapse at the end.

Significant coefficients by level in Quantile regression. The weight of Run8 increases as the layer slows down.
Figure 4. Effective elements change depending on the level. In the Sub-100 zone, preventing collapse in the final stage is the top priority.

In actual combat, it is effective to tackle in the following order.

  1. Zero walking in Run7-8: Prioritize continuity over speed.
  2. Respiratory reconstruction 30 seconds after the event: Stabilizes the entry of the next run.
  3. Final stage Wall Balls split fixed: Stop the stall chain.

The goal is not the ``fastest lap,'' but rather a ``lap composition that will not collapse until the end.''

4. 4 week plan (90 minute wall countermeasures)

Expect 2-3 HYROX-specific sessions per week. The goal is to reduce late-game stalls and increase reproducibility.

Week 1: Visualizing the current situation

Week 2: Create a stable distribution

Week 3: Reproducing under fatigue

Week 4: Check with race specifications

5. Race day running rules

  1. Runs 1-2 are not "earning sections" but rather preparation sections for the final stage.
  2. Don't raise it too much in Run3-4. If breathing becomes irregular, return to normal position immediately.
  3. Run 7 will start with a 200m limit, and Run 8 will aim for zero walking.
  4. Wall Balls lowers the failure rate by splitting and fixing rather than completing consecutively.
  5. In the final stage, prioritize your own reproduction rules, not those around you.

6. Common mistakes and how to fix them

failure pattern The scene of waking up How to fix
Overpacing in the early stages Run1-4 Enter Run 1 a little later than the target and match it with Run 2
Walking starts in Run8 final stage Hold down the 200m at the beginning of Run 7 and fix your breathing in Run 8.
No reproducibility of stall whole lace Record Run8/Run1 ratio and walking seconds each time to manage trends
Collapse with Wall Balls final stage Fix split reps and decide which rep to restart in case of failure

7. Three numbers to track with HYFIT

If these three things are improved at the same time, the reproducibility of time reduction will increase in the Sub-100 band. On the other hand, if any of them is deteriorating, it is safer to correct the distribution before adding load.

8. Frequently asked questions

Q. Will I be able to break 90 minutes if I just improve my running ability?

Running power is necessary, but in the Sub-100 range, it is more effective in the short term to stop the collapse in the final stages. It is effective to first prepare the distribution and stall management, and then build up running power.

Q. How can I use Quantile results in practice?

Prioritize elements with high coefficients in layers close to your band. For the Sub-100 band, it is highly effective to prepare for Run8 collapse suppression first.

Q. How much will change in 4 weeks?

Although there are individual differences, athletes who can reduce the number of seconds they walk and the number of stops can often aim to shorten their time by 1 to 3 minutes.

Conclusion of this article

Mini terminology explanation

Notes on method and interpretation

The numbers in this article are based on statistical analysis of public race results. Although it does not directly prove individual cause and effect, It is quite effective for practical use in determining practice priorities.

Data source

Yamanoi, S. (2026). Winning Strategies in HYROX: A Machine Learning Approach to Race Performance Optimization. Zenodo. DOI: 10.5281/zenodo.18683662

Record: https://zenodo.org/records/18683662

SY

Author profile

Shuta Yamanoi

Writes strategy articles for practitioners based on research on HYROX data analysis (58,852 people, 58 tournaments).

View detailed profile

Turn your analysis into your next race result

By recording each section with HYFIT, you can track the effectiveness of your 90-minute wall preparation until the next race.