1.3 Million HYROX Results: Pacing, Progress, and the Podium

01 Eight seasons: more races, bigger fields, and returning athletes

Notes: participations count Singles=1, Doubles=2, Relay=4 (finishers, not registrations). * Unique athletes are matched approximately by name+nationality+sex (Open/PRO Singles only), so small errors are possible; early seasons with incomplete identity data show "—". † Season 3 (2020–21) was COVID-disrupted: only four US events — a structural break, excluded from YoY.

The table in one sentence: HYROX is on a steep growth curve — participations grew ~54× from S4 to S8 (roughly 1.7× per year), still expanding at +157% in the latest season. Two engines fire at once: more races (24→111 events) and bigger races (928→10,793 participations per event), see Fig. 1. The structure is being rewritten too: EMEA still holds ~60%, but APAC has surged from 3.7% to ~20%, the fastest-expanding region (Fig. 2); the female share rose from 39.6% to 45.9%, near balance (Fig. 3) — HYROX is turning from a Europe-centric, male-majority event into a global, gender-balanced mass sport.

Age: who races, and how fast each age needs to be to podium

• Composition: ages 25–44 carry the field (~three quarters) — 30–34 largest (23.4%), then 25–29 (20.6%) and 35–39 (19.0%); over-55s are under 4%.
• Podium line vs field median: podium finishes run ~20–27 minutes faster than the age-group median — ~26–27 min in the prime bands (Open men 30–34: podium 60.0 vs median 85.8) narrowing to ~20 min after 55. PRO field medians beat Open by 7–9 min at matched ages, yet the podium lines are similar — Open's far bigger fields run deep at the top.
• The podium bar rises every year: same-city, same-age-group third-place cutoffs got a median 1:57 faster S7→S8 — ~1.5 min in Open, ~3 min in the more contested PRO. Staying on the podium means getting ~2 minutes faster per year.

Podium = top 3 of each event × division × sex × age-group cell with ≥10 finishers; the Fig. 5 band is the middle 50% of finishers.

Retention: do athletes race again within 3 / 6 / 12 months?

Starting at each athlete’s first Open/PRO Singles race in a season, we ask whether they race again and how often returners race. Identity and date conventions: see footnote².

In-season frequency is rising too: Open/PRO singles athletes average 1.28 (S6) → 1.30 (S7) → 1.35 (S8) races per season, with the share racing twice or more up from 19.9% to 22.8% — repeat racing keeps growing.

02 Target finish → running pace and station budget

Quick reference: effective pace by target finish

Effective pace = (eight runs + RoxZone) / 8.7 km — an empirical approximation of course length; there is no official uniform distance. Relay rows are team results (four athletes taking turns) — don't benchmark your personal pace against them.

* Insufficient sample (n<100).

03 Where does a five-minute gap come from?

We split finishers into five-minute bands and compare each band with the next (T+5 vs T), decomposing the mean difference across Running, RoxZone and the eight stations — where do 75-minute athletes save their five minutes over 80-minute athletes?

Run the same five-minute decomposition across the three formats and the difference is visible at a glance — singles have five components above 5%, doubles collapse to running plus RoxZone, and relay returns to a singles-like structure:

04 Pacing: your type, the front-runners, and where gaps open

4.1Which pace type are you?

4.2What do the fastest look like?

Three facts worth remembering

• The top does not belong to the extremes. 61.3% of the top 5% are balanced (both tilts within ±155s), runner 19.8%, strength 15.5%; fast-start slow-finish just 3.5%. The faster the athlete, the less lopsided the profile.
• Balanced athletes reach the top 5% at 2.4–2.7× the rate of the tilted types (10.2% vs 4.2/3.8%) with the fastest median (82.0 min). No weakness is the strongest build.
• The other trait at the front is the negative split: 64.5% of the top 5% run faster late than early (F<−1%; 41% field-wide), rising monotonically with ability. The blunt fader name stays earned — 0.6% conversion.

How each type gets faster: answers from same-type athletes who made the top 5%

Comparing each type's top-5% members with the rest of their own type, all four share one direction — regression toward the middle: you advance by fixing the weakness, not stretching the strength:

• Runner: top-5% runners carry 68s less station debt (Wall Balls −28s) while holding a milder run edge (+37s) — train the big four, don't stack more mileage.
• Strength: top-5% strength athletes run 42s faster (−5.3 s/km) with a slightly smaller station edge — build the aerobic base and running economy.
• Balanced: top-5% balanced athletes are slightly better everywhere (stations −14s), no single magic dial — squeeze RoxZone and Wall Balls.
• Fast-start slow-finish: the few who advance fade 2.7pp less — the cheapest gain is pacing discipline itself: hold Run 1, and train "running while tired" as a capability.

Two common questions, answered directly

• "Attack Run 1?" The fastest athletes run fast Run 1s in absolute terms yet allocate them conservatively (Top-100: 2.3–3.9pp relatively slower, 2.8–4.3pp less fade). PB-day data splits it in two: Run 1 is 11.3s faster on PB day — "slightly faster" is right; but it improves least of the eight runs (Run 8: 18.8s), with gains concentrated in mid-to-late retention. Restricting to races 30/45 days apart doesn't change the direction.
• "How do PBs happen?" Not by secret tactics: of a mean 305s PB gap, Running is 132s (43%), Wall Balls 42s, BBJ 28s, Sled Pull 27s, RoxZone 23s; Ski+Row just 10s. On a good day nearly everything is faster — and the mid-to-late race, wall balls and loaded stations separate good days better than the first kilometre.

4.3Where gaps open — and how singles, doubles and relay differ

Station strategy: the data's "big four" and "small four"

"Which station separates athletes" is a computable question: split the spread in finish times exactly into each component's contribution share (a variance decomposition). For Open singles the answer is tidy — running 43.9%, RoxZone 11.1%, then the big four: Wall Balls 12.9%, Burpee Broad Jump 9.7%, Sandbag Lunges 7.9%, Sled Pull 6.0% — 36.4% combined; the small four (Sled Push 3.0%, Farmers Carry 2.1%, Row 2.0%, SkiErg 1.5%) total just 8.6%. Four times the separating power — not hand-picked, computed, and stable across the fast, middle and slow thirds of the field.

Wall Balls: not the longest component, but highly discriminating

Wall Balls often contributes 10%–13% of a five-minute gap while consuming roughly 6%–8% of race time, so time-normalized leverage often exceeds 1.4. Within-athlete PB and podium comparisons agree. Brandt’s n=11 physiology study is only external context: Wall Balls showed peak HR/lactate/RPE.

RoxZone: real transition time

The pace calculator, five-minute decomposition, cross-season within-athlete pairs and PB decomposition all use the official RoxZone transition time. Open men 80→75 differ by ~32 seconds in RoxZone (11.0%, leverage 1.32). Top-10% repeat athletes are 23 seconds faster on PB days; within-athlete podium races are 25 seconds faster than non-podium races. RoxZone is real, visible time; whether to walk or jog it, result data can't answer — there is no per-transition or heart-rate data.

Ski / Row: small direct budget does not prove protective slowing

Ski and Row together usually explain only ~3%–5% of a five-minute gap, so they are not where the seconds are. But "give 5–10 seconds on the erg to protect the next run" doesn't check out in this data: athletes who pace ergs conservatively differ in ability to begin with, and our dedicated test failed. Proving it would take pre-race erg benchmarks and a randomized crossover trial.

Does the focus shift across levels?

Decomposing each five-minute step from 100→95 down to 60→55 (all cohorts agree in direction; use the tool above to inspect any step): Running is the top budget throughout; Wall Balls and RoxZone hold high shares at almost every stage; Sled Pull and Lunges grow near 60 minutes — loaded stations matter more as you get faster; Ski/Row stay last everywhere.

And in absolute time: the station-pace heatmap by finish band

Note: §3 and the text above decompose shares of a five-minute gap; this heatmap shows absolute time spent per component at each finish band (same data as the §2 calculator; cohort selectable).

How to read it: each row is one component, each column a target finish band (55–100 min); cells show the band's median time; each row's hue is its category (amber = running, blue = RoxZone, vermillion = core four, grey = other stations, consistent site-wide), and shading within the row tracks growth across bands. Read across a row for "how this component changes from 80 to 75 minutes"; read down a column for that band's full station budget (same data as the calculator). The Running row shows pace large, total time small.

The “wall” is not one station—or just running

Late-run fade and late-station deterioration move together at population level and remain positively associated within athlete, but single-race directional accuracy is only 53.3%. Heavy-fade races show relatively faster Row but worse BBJ/Lunges/Wall Balls—consistent with resting on self-paced ergs and expressing fatigue later. One race can't diagnose "you hit the wall"; the signal only means something across races.

Singles, Doubles, Relay: different races, different levers

Run the same variance decomposition on all three formats: in Doubles, running's share of differentiation rises to 58.8% (PRO Doubles 55.0%) while the big four collapse — Wall Balls 12.9%→6.3%, BBJ 9.7%→5.9%, Lunges 7.9%→5.4%; Relay matches singles almost component-for-component (running 43.3%, WB 12.6%). The mechanisms separate cleanly: in doubles, station times vary less between athletes (Wall Balls narrows to 0.70 of the singles spread) and decouple from running fatigue — the fingerprint of work-sharing; relay stations don't narrow (they widen) yet also decouple — the fingerprint of freshness.

• RoxZone share differs by format: singles 11.1%, doubles 12.6%, relay 9.1%. Official timing has no separate changeover clock, so we report the numbers without attributing them.
• The last run can be sprinted: in course-matched pairs, doubles run Run 8 1.3–2.6pp less slowed vs their own early race — the other side of the sharing mechanism (the final wall balls are shared).
• Training implication: doubles prep shifts priority to running and shared pacing, discounting the big-four logic; relay prep follows singles logic — its edge is freshness and leg assignment.

Format note: doubles station logic is different

At the same 80-minute finish, Doubles spend 5–6 more minutes running (shared pace, slower-partner constrained) and buy it back at shareable stations: BBJ −1.4/−1.5, Wall Balls −1.1/−1.7, sleds ~−1 each, Lunges −0.8/−1.0 min; ergs barely help (Ski −0.3, Row −0.1) and RoxZone costs +0.5–0.8 min (no separate changeover clock in official timing, so we do not attribute it). Doubles priorities therefore tilt toward running and shared pacing; the singles "core four" logic carries less weight. Relay allocation is nearly identical to singles (every station |Δ|<0.5 min) — relay's edge is freshness, not sharing.

05 Converting Open, PRO, Doubles and four-person Relay

Conversions use same-athlete, same-team, same-event pairs. These athletes skew strong and race often, so treat the numbers as magnitudes.

Method: same athlete, two races within 45 days; times are converted to event-relative position then mapped back to minutes, killing venue effects. Widening the window to 45/90/180 days shifts results by ≤0.15 min, and strict same-event pairs agree. Relay uses teams where at least three of the four members could be identified (this adds virtually no bias — within 0.1 min of the 270 fully identified teams).

The Open→PRO cost is load-specific: Sled Push +1:20, Pull +1:33, Wall Balls +1:28, Lunges +0:52; Ski/Row/BBJ near zero. Doubles has no fixed benefit — the gain depends entirely on the partner gap; see the table below.

How the partner gap changes the outcome

Close partners both gain from sharing; past a 20-minute gap the faster partner is bound by the shared run and the team underperforms their singles time. The windowed cohort (7,818 teams) reproduces the shape: the >20-min cell sits at −0.6s ≈ 0.

Conversion varies by ability

Reading straight off the quick-reference: at 60 minutes Doubles run ~19 s/km faster effective pace than Open men, but ~54 s/km faster at 90 minutes — the slower the band, the larger the benefit of sharing stations (stations weigh more in slower athletes' totals). The PRO-Doubles gap also widens down the field. Partner gap remains the single biggest variable (the +0:55 reversal above).

The same athlete across three formats: a conversion ladder

Among the 66 relay teams where all four members' identities could be matched: relay beats the four members' singles mean by 15:04 [confidence interval 14:02–16:40, likewise below] and the fastest member by 7:30. Cleaner still, the three-format sample: 1,382 athletes raced singles, doubles and relay within the season (967 all three at one event) — vs their own singles, doubles is 7.9 min faster [7.4–8.4] and relay 10.4 min [9.8–10.9]. These triple-format athletes are notably stronger and keener; treat these as magnitudes.

06 How to podium: does podium racing differ from ordinary PB racing?

Podium means top three in each Season-8 event × Open/PRO × sex × age-group cell with n≥10: 8,562 results across 2,854 cells. We compare within cell, then within 3,343 athletes who have both podium and non-podium races.

The key test: a different strategy, or just faster?

The table above has a trap: podium athletes are faster than their cells, so "strategy differences" may just shadow ability. The ability-matched version pairs each podium athlete with a top-10% non-podium athlete of nearly identical finish time (within 2%) in the same cell and compares shapes only (488/609/75/273 pairs; confidence intervals resampled event-by-event so athletes from the same race aren't treated as independent).

• Every shape difference collapses to null. Matched Run-1 diffs are −0.13 to +0.40pp and fade CIs all span zero (lone marginal exception: Pro women, −1.0pp, n=75). Podium racing is not a different pacing strategy — it is the same strategy executed on more ability.
• "Podiums require gambling on Run 1" is refuted. Podium finishers show less Run-1 variance (standard deviation: Open men 4.98 vs 5.72, non-overlapping confidence intervals) and only 39–42% sit on the fast-start side of the contrast median (50% baseline) — the podium population skews controlled, not risk-taking.
• What survives matching is small but real: transitions ~3–4 s faster per race (significant for Open/Pro men) and second-order station tilt (relatively easier ergs, 1.8–2.9% faster Wall Balls/Lunges in some strata) — podium athletes win on clean process, not aggressive tactics.
• Formats: Doubles reproduce the same pattern (null matched shapes; transitions −3.3 s); Relay's 16 splits are four different legs, so "pacing shape" would measure roster order, not pacing — explicitly skipped.

Practical translation: there is no podium-specific pacing chart. Execute the controlled start your ability supports, claw back 3–4 s in transitions, build a relative edge in Wall Balls/Lunges — and accept that podium outcomes are decided mostly by the ability gap to your cell's top three, then by clean execution. One blind spot: runaway winners cannot be matched, so this covers contested podiums, not dominant solo wins.

07 Finally: seven headline findings

1. Eight years: from a handful of races to over a hundred per season. HYROX grew from 9 events in S1 (2019) to 111 in S8 (2026), with ~1.2M participations in a single season and 2M+ cumulatively; APAC's share rose from 3.7% to ~20%. Stickiness is rising too: over 40% race again within 12 months (41.7% full-S7; 43.6% early-S8 cohort¹), and races per athlete per season climbed from 1.28 to 1.35.
2. Your target finish maps directly to a pace. Open men finishing in 60 / 70 / 80 minutes hold an effective pace (8.7-km incl. RoxZone) of roughly 3:57 / 4:32 / 5:09 per km (Open women ~3–6s slower at the same band); PRO needs another 5–16 s/km, Doubles can run 20–45 s/km slower, and Relay is within 2–4 s/km of singles. Full bands in the §2 calculator.
3. Five minutes faster: singles train running + the big four; doubles bet on running. For singles (60–80 min bands) running drives ~41% of the gap and the core four — Wall Balls, Burpee Broad Jump, Sled Pull, Sandbag Lunges — ~36% combined; Ski + Row total just 4.8%, the worst seconds-per-effort; RoxZone is 10–11% — run the transitions. Relay follows the same law (running 45%, core four 32%). Doubles are a different race: running 60%, RoxZone 11.7%, and no single station above 5.5% — train the shared run, not station splits.
4. Four pace types; simply not fading beats a quarter of the field. Typed by two computable indices (balance B, fade F): runner 23%, strength 20%, balanced 30%, fast-start slow-finish 27%. Runner, strength and balanced athletes all reach podiums; the fader almost never does (0.6% conversion; median finish 14–21 min slower) — just not fading puts you ahead of ~27% of the field. One more trait at the front: 64.5% of the top 5% negative-split the race. Search your race in the results lookup to see your type.
5. Format conversion rules of thumb (within-athlete, 45-day window). Open→PRO: ~6.8 min slower for men, ~4.9 for women (almost all at the loaded stations); a pair racing Doubles beats their singles average by ~11 min, but it hinges on the partner gap — ≤5 min apart the team still beats the faster partner by 8:40, past 20 min the edge vanishes; Relay beats the four-member average by ~16 min (n=1,016 teams).
6. Podium strategy is no different from elite strategy — podium finishers are simply faster at every station. Matched to equally-fast top-10% athletes, their pacing shapes coincide; their increment over the top 10% concentrates in the core four (~10–11% each) and RoxZone (9.7%), with running just 6.4%. There is no podium pacing chart: executing your level cleanly is the podium strategy.
7. Judge yourself within your age group. About three quarters of the field is 25–44; the same 90-minute finish is top 61% at 25–29 but top 27% at 60–64. At matched ages PRO medians beat Open by 7–9 minutes — division choice is itself a selection.

注Footnotes

1. At the data freeze, only the earliest 6,773 S8 entrants had completed the full 12-month window (later entrants hadn't had 12 months yet); a selective sample, not directly comparable with S7.
2. Retention starts at each athlete's first Open/PRO Singles race of a season; identities are approximate name+nationality+sex links, with observation windows ending at the freeze. S1–S5 lack joint date and identity coverage.

附Appendices