Methodology

This is a data site, so you should be able to check our working. This page says where every number comes from, what is measured and what is modelled, and where we are currently uncertain.

Coverage is incomplete, and we say so on the page

We do not hold every finisher of every race. For most published divisions we hold a partial sample rather than the full field, and those leaderboards carry a “partial results” notice stating how many rows we have against how many the race recorded. A rank shown on this site is the athlete’s rank in the actual race, not their position within our sample.

Modelled vs. measured

We label these differently everywhere on the site, because conflating them is how data sites lose trust.

Modelled: Derived from a documented assumption rather than observed results. The pace calculator and the “typical splits” tables on station guides are both modelled. They are useful for planning, but they are not evidence of what any real field actually ran.
Measured: Taken from actual recorded results. Finish times, ranks, age-group placings and the station-by-station splits on result pages are all measured — they are the values the timing system recorded, not estimates.

How the pace calculator works

A HYROX race is 8 km of running alternating with eight stations, plus roxzone — the timed transition between the run lane and each station. The calculator distributes your goal finish time across all 24 of those segments.

Rather than assuming a shape, we fit one. For each category and division we regress every station’s race time against run pace per kilometre across our own results — 1,937 open men’s races, 1,034 open women’s, and similar samples for Pro, Doubles, Pro Doubles and Relay. Solving that relationship for your goal time gives the pace it implies, and every split follows from it.

We report the fit quality, not just the number. Some stations are predicted well by running fitness; others barely at all. Sled push has an R² of 0.21 for men and 0.11 for women, because it depends on strength and the venue floor rather than aerobic fitness. Those stations are marked as indicative in the calculator rather than presented with the same authority as a rowing estimate.

Individual segment targets are derived from differences of rounded cumulative times, so the plan always sums to exactly your goal rather than drifting a few seconds off through rounding.

Known limitation: our dataset over-represents faster athletes, so the mean run pace in the fit is quicker than a typical race field and predictions for slower athletes will tend to be optimistic. The finish-time distributions and athlete counts shown alongside are drawn from the same dataset and carry the same caveat.

The finish-time predictor additionally converts fresh, max-effort station tests into race-day equivalents using fatigue multipliers. We hold race splits but not paired fresh tests, so that conversion remains modelled rather than measured — it is the one estimated component left in the tools.

Where split ranges come from

The Elite / Competitive / Age-group / First-timer bands on each station guide are computed from 18,930 of our own results that carry full station splits — pooled across every category and gender, since the guides themselves are generic. A tier is a quartile of overall finish time (Elite is the fastest 10%, down to First-timer at the slowest 25%); the range shown is the 25th–75th percentile of that station’s actual recorded time within the tier.

That is a real, if pooled, distribution — not a per-division breakdown. We have not yet split it by category, gender or season; the sample is large enough to be meaningful in aggregate, but a Pro women’s sled push and an Open men’s will both be folded into the same band above.

Where the data comes from

Results are compiled from publicly published race records. HYROX races are timed by an official timing provider, and the official record is the authoritative one — where our figures disagree with it, the official record is correct.

Coverage varies by event and division for reasons that have nothing to do with the race itself: some divisions are captured in full, others only in part. We show what we hold and label it honestly rather than presenting a sample as a complete field.

Things that make splits hard to compare

Venue surface. Sled push and pull times vary substantially between carpet, turf and different sleds. The same athlete can post very different sled splits at two events for reasons that have nothing to do with fitness.
Division standards change. Loading and rep counts have been revised between seasons, so a split from 2022 is not always directly comparable to one from 2026.
Roxzone length differs. Venue layout changes how far you walk between the run lane and each station, and that time is on your clock.
Field size affects rank. A rank of 40th means something different in a 200-person field than in a 2,000-person one, which is why we show field size alongside rank.

Corrections

If a number on this site is wrong, we want to fix it. Athletes can request a correction or removal of their own data via the contact page. We will correct verified errors and note substantive corrections on the page where they occurred.

Independence

HYROX Registry is an independent project. It is not affiliated with, endorsed by or connected to HYROX or Upsolut Sports AG, and we use the HYROX name only to describe the sport this site covers. The official race organiser publishes its own results, and where our figures disagree with the official record, the official record is correct.