SYNTH-49 · Guide
XCO is raced as repeated surges above MAP on a near-threshold floor, and cyclocross is assumed to be by transfer (no cyclocross power data)
Train the aerobic ceiling and repeatable supra-maximal power together, keep heavy lifting through the season, and count the weekend races as the week's intensity
The question it answers
How to prepare for an off-road race of roughly one to one and a half hours that is decided by repeated short surges: what the race asks for, which two qualities to build, where strength and skills fit, how to handle the start, and what a week looks like when there is a race every weekend. VO2max interval shape is SYNTH-33, sprint formats are SYNTH-46, lifting is SYNTH-43; none of them is restated here. Marathon MTB (XCM, 3-6 h) is not covered.
What the race asks for (as the base states it)
| Measure | XCO | Cyclocross |
|---|---|---|
| Duration, intensity | ~80-100 min, mean HR close to 90% HRmax (REF-742, A, SR of 53 studies); 87% VO2max (REF-741, B, 16 national/international riders); mean 283 W = 68% of MAP at 91% HRmax (8 elite men, 30 international race files, REF-897, B); 96 min at a mean 76% of CP (7 national riders, 2 women, REF-898, B) | 93.6% of race time above the HR at 4 mmol/L, mean HR 170.8 vs race maximum 177.8 bpm, post-race lactate 8.3 mmol/L (REF-740, C+, n = 8, HR and lactate only, no power) |
| Time above MAP | about a quarter of the race (REF-742); 25% above MAP and 37.0% above VT2 (REF-741); 26% above MAP and 25% below 10% of MAP (REF-897); 40% above CP (REF-898) | not measured — no cyclocross study met REF-742's criteria |
| Surges | ~3-10 s at ~130-140% MAP, ~15-20 per lap in elite riders (REF-742); counts depend on the definition: ~18 per lap of ~40 s (Granier) vs 334 per race of 4.3 s, 10.9 s apart (Prinz), both as cited inside REF-742; most bursts 5-30 s (REF-741); bouts above CP ~8 s at ~120% of CP (REF-898) | not measured |
| Pacing shape | fast start, speed stable or slightly falling later (REF-742); first lap very high, later laps more aerobic (REF-741); power fell from the start loop to lap 1, then held (REF-897); by the last lap, bouts above CP kept their number and length but lost 25% of magnitude, and W′ spent per lap fell 45% (REF-898) | no difference between laps in lap time or HR (REF-740) |
| Between surges | technical riding keeps oxygen uptake high (REF-742) | no low-intensity time (0.4% below 2 mmol/L HR, REF-740) |
Two reading rules follow.
Read an off-road file by the count and length of efforts above MAP and time above MAP, not by average or normalised power (the entry's practical reading, not a tested rule); zone time by %HRmax and by %MAP did not track zone time by %VO2max (r = −0.01 and −0.02, full text) (REF-741, B); a mean of 68% of MAP (REF-897) or 76% of CP (REF-898) hides a quarter to two-fifths of the race above those ceilings.
In cyclocross, heart rate sits near its race maximum from the first minute, so it cannot pace the race (REF-740, C+).
Default (what to prescribe first)
For a Tier 2-4 rider with a consistent base (if there is no base, the lever is consistency first — SYNTH-42 gate):
Build two pillars at once, not one. An aerobic ceiling (MAP/VO2max) AND repeatable supra-maximal power; both moved race-relevant markers in XCO riders and HIIT and SIT were roughly equivalent head-to-head (REF-742, A, Tier 3-5 XCO riders). Each was tested on its own; no trial compares training both against training one, so "at once" is that entry's practical reading, not a tested result. Ceiling sessions: shape per SYNTH-33. Supra-MAP sessions: format per SYNTH-46.
Heavy strength in preparation, never zero in season — the burst-event row of SYNTH-43 (default plus a unilateral lift).
Decision (who → what)
| Situation | What to do | Source and strength |
|---|---|---|
| XCO rider, preparation phase | Two pillars. Tested doses in XCO riders: 30-s all-out SIT, 2×/week, 4-6 × [4 × 30 s : 90 s], added to normal training for 8 months → VO2max +6%, MAP +5%, sprint power +4%; 6 weeks of HIIT (7-10 × 4-6 min) vs SIT (8-12 × 30 s) → race time −5% vs −3%, peak power +8% vs +5%; 4 weeks of 20 s @ 115% MAP / 40 s @ 50% MAP, 60-90 reps, 2×/week → TT −4%, MAP +8%, power at 4 mmol/L +13% | REF-742, A (direct race-performance assessments of any intervention are scarce; participant description often poorly reported) |
| Rider whose files show the surge count or size collapsing in later laps beyond the usual fade while the ceiling tests well | Weight the block toward repeatable supra-MAP work (SYNTH-46 formats), keep one ceiling session | Inference from REF-742's two-pillar reading and REF-741's file-reading rule. The usual fade in 7 national riders: same number and length of bouts, 25% smaller by the last lap (REF-898, B). No trial compares this choice |
| Rider with a low measured ceiling (Tier 1-3) | Ceiling first (SYNTH-33), supra-MAP work as a small standing dose | SYNTH-42 (work above threshold is the starting bet for most amateurs; practice); oxidative supply is one of three energy limits in repeated sprints (REF-551, B+, narrative review, team-sport-led) |
| Any XCO/CX racer, strength | SYNTH-43 burst-event row; keep ~1 heavy session every 7-10 days in season (dose is practice) | Heavy lifting raised anaerobic power ES 0.56, efficiency ES 0.35, not VO2max (REF-743, A, 17 trials, 262 cyclists, low certainty); in elite XCO men, sprint power +16% after one preparation period and up to +25% (leg press +44%) over 3.5 years (REF-047, B, case report, 2-4 riders); stopping for 8 in-season weeks cut squat jump, Wmax and Wingate power vs controls (REF-744, B, 7 per group, elite men) |
| Cyclocross racer | The XCO structure, shifted toward a threshold floor with surges on top | Transfer from XCO (REF-742, REF-741) — the base holds no cyclocross training study; the threshold-floor framing is REF-740's application (C+, n = 8) |
| Tier 1-2 amateur, or a woman | The same shape, with the numbers read as elite-male reference values, not targets | REF-742 limitations: Tier 1-2 and female riders under-represented (Tier-5 women n = 10); REF-741: W/kg will not transfer to amateurs, the pattern will |
The start and the warm-up
The grid shapes the result. In World Cup XCO 1997-2007, start and finish position correlated in every season (P < 0.01): on average 57.4% of men moved fewer than 15 places and 62.9% of women fewer than 10 (REF-872, B, observational; the fastest riders are also gridded in front). Not from a study; the base's own practice: count the points that earn a better grid as part of the season plan.
Prime with a sustained bout, not with sprints. A moderate or heavy (below-CP) bout finished 10 min before a 7-min trial raised mean power ~2-3%; repeated all-out sprints did not (REF-270, B, n = 12 well-trained cyclists). All-out 5 × 10 s cost −5.8% on a 3-km TT (REF-272, B, n = 10).A ≥50-min warm-up with four sprints lowered Wingate peak power versus 15 min with one sprint (REF-276, B, track cyclists). Transfer: all three tested efforts of 30 s to 7 min in the lab; none tested a one-hour off-road race.
Severe-intensity priming plus a 12-s all-out start made a ~4-km TT ~9 s faster (402 vs 411 s); either alone did not (REF-279, B, n = 9 men, lab). The priming there was severe-intensity (above CP), not the below-CP bout of REF-270; the two were never compared. Inference: the closest tested analogue of a mass-start off-road opening, not a test of it.
Keep the legs actively warm on the grid. After a 30-min wait, electrically heated pants raised 30-s sprint peak power 9.6% vs tracksuit pants; insulation alone did not hold muscle temperature (REF-745, B+, n = 11 trained men, lab). The effect on the rest of the race was not measured.
Not from a study; the base's own practice: rehearse the start (from a standstill, all-out for the length of the start straight, then settle into a hard first lap) once a week in the last weeks before and during the season.
Skills, technique and cyclocross running
No entry in this base measures the time cost of cornering, descending, dismounts or running, or trains any of them. What the base does say: technical riding in XCO keeps oxygen uptake high (REF-742), so technical sections are not recovery.
Not from a study; the base's own practice: one skills session a week (cornering, lines, and for cyclocross dismount/remount and short run-ups), done fresh and at low physiological cost, and skills inside the surge session on a real loop in the specific phase.
Running for a cyclocross rider: no base entry quantifies it. A narrative review notes that cyclists often have lower bone density than ball- and power-sport athletes, and its entry advises progressing volume slowly when running is added to a non-impact background (REF-522, C+). Not from a study; the base's own practice: short run-ups a few times a week, never a running volume block.
Weekly structure in season (racing every weekend)
Races are the high-intensity sessions. Coaches of world-class athletes report that competitions supply a large share of the year's HIT; counting them against the HIT budget rather than adding intervals on top is that entry's practical reading (REF-302, B, qualitative interviews, Tier 4-5; transfer to amateurs is inference).
Not from a study; the base's own practice, for one race a week: day after the race easy or off; one midweek quality session, alternating ceiling (SYNTH-33) and supra-MAP (SYNTH-46); one heavy lift at least 48 h from the race (the 48 h is practice in REF-744's entry) and not in the 24 h before the quality session (SYNTH-31); one skills session; a short opener the day before. For two races a weekend: drop the midweek VO2max-type session, keep the lift and the skills.
Between two race days, refuel as a dose: 1-1.2 g·kg⁻¹·h⁻¹ carbohydrate early, 0.3-0.4 g·kg⁻¹ protein, 125-150% of a measured fluid loss with sodium (REF-748, A, UCI consensus written for elite riders). In-race fuelling: the under-90-min row of SYNTH-02 (that row is consensus, REF-748; an 80-100 min race sits on its upper edge).
Post-race damage markers are a load signature, not a fault: after a real race, myoglobin rose in U23 mountain bikers and CK did not rise in the mountain bikers as a group, but the better-placed riders showed the larger CK change (REF-703, B, n = 11 MTB men). Not from a study; the base's own practice: judge recovery by readiness, not by one marker.
How to progress / periodise
Of the XCO trials REF-742's entry lists, the longest ran SIT 2×/week for 8 months added to normal training; the others ran 4 and 6 weeks.
When NOT to use it / red flags
A rider without a consistent base → SYNTH-42 gate first; surge blocks before a base are false precision (inference).
Do not judge an off-road race from NP, average power or HR zones (REF-741).
Do not build only one pillar (inference): REF-742 shows each works and that HIIT and SIT were roughly equivalent head-to-head; that neither covers the other was not tested.
Do not stop lifting at the first race (REF-744).
Do not add a VO2max session on top of a double-race weekend (practice, from REF-302's budget rule).
What the evidence does and does not show
Shows: XCO demand profile and the effectiveness of HIIT, SIT, 20/40 and strength on race-relevant markers in XCO riders (REF-742, A); the surge-based file reading (REF-741, B); race power against MAP and CP and the late-lap shrinking of surges (REF-897, REF-898, B); the cyclocross HR/lactate profile in eight riders (REF-740, C+).
Does not show: much about race results themselves — direct race-performance assessments are scarce (REF-742); any cyclocross power data or cyclocross training trial; anything on skills, running, or the in-season week; start position only as a World Cup association (REF-872). Everything in those rows is labelled practice or transfer.
Evidence gaps
Cyclocross power demands (surge count, time above MAP, coasting/running share) — no primary in the base; no full-text candidate identified.
Running and dismount cost in cyclocross — none.
Start position in amateur or modern-format XCO — REF-872 is World Cup 1997-2007 and associational.
Skill training (cornering, descending) and its time value — none.
Tier 1-2 amateurs and women in XCO — under-represented in REF-742.
How this entry was checked and changed
Content correction 2026-09-27
Revision 2026-09-29: race-demand table, file-reading rule and the late-lap-collapse row now use REF-897 (Granier 2018, B, 8 elite men: 68% of MAP mean, 26% of time above MAP) and REF-898 (Næss 2021, B, 7 national riders: 40% of time above CP; late bouts 25% smaller, same count); REF-872 (Macdermid 2012, B: World Cup start position tracks finish) opens the start section; XCM gap points to REF-859. Default and level unchanged.
Verification 2026-10-01: all 22 cited REF entries opened in the KB.md snapshot; every number in the race-demand table, the decision table, the start/warm-up section and the recovery line matches its entry; every level/design/n written next to a refId matches the entry head block (no LEVEL findings).
Verification 2026-10-01: no SOURCE flag by the spec's rule (all cited REFs carry a DOI). For information: REF-897, REF-898, REF-872, REF-859 are 'written from the abstract only'; REF-742's numbers 'were not re-checked against the full text on 2026-09-27'; REF-740's full text is no longer served.
Verification 2026-10-01: Level line B is honest for the XCO demand description and says itself that cyclocross is C+ plus transfer and the week/skills/start rehearsal are practice; no LEVEL-OVERALL finding. Of 9 Default/decision rows only 2 are direct evidence (tested XCO doses; elite-male reference values), 4 are inference/transfer/practice, 3 mixed.
Verification 2026-10-01: cross-framework checks passed for SYNTH-43 (burst-event row = default plus a unilateral lift; 1 session every 7-10 days in season), SYNTH-42 (engine default at Tier 1-3, labelled inference there), SYNTH-46 (4-8 weeks ending 2-3 weeks before the A-race, house practice), SYNTH-31 (24 h rule, house practice).
2026-10-02: aligned with the rewritten SYNTH-42 (missing stimulus, one lever per block, level C+): "engine is the default lever at Tier 1-3; inference over cross-sectional data" -> "work above threshold is the starting bet for most amateurs; practice"; "the limiter is consistency" -> "the lever is consistency"; cross-reference label.