cycling-training/references/road-racing-demands.md

Road-racing demands (mass-start: road races & criteriums)

Load this doc when: planning for a mass-start road race or criterium — any bunch event decided by positioning, surges, and a finish, rather than a steady solo effort.

Data/knowledge split: MCP gives you the power streams, record power profile, and load. This doc explains the demand profile — why road racing is stochastic, what wins it, and how to train for it.

Contrast with the other demand docs: a road race is the near-opposite of tt-demands.md. It is intermittent, stochastic, tactical, and drafting-dominated — structurally closer to XCO (mtb-xco-demands.md) than to a TT. Don't pace a road race like a TT; the winning move is a short supra-threshold effort off a variable base, not an even wattage.

---

The power profile: low mean, high variance, decisive surges

Mass-start road racing has a stochastic power-output profile — long stretches at modest power (sheltered in the bunch) punctuated by repeated short efforts above maximal aerobic power.

• Across 148 Tour de France mass-start stages, mean power was only ~3.1–3.3 W/kg (218–234 W) despite the racing being decided by far higher climb and attack efforts (Vogt et al. 2007, Int J Sports Med 28(9):756–761; PMID 17497569; DOI 10.1055/s-2007-964982).
• Professional tour racing showed relatively low mean power interspersed with multiple high-intensity surges above maximal aerobic power, varying by race type (Ebert et al. 2006, Int J Sports Physiol Perform 1(4):324–335; PMID 19124890; DOI 10.1123/ijspp.1.4.324; women's World Cup: Ebert et al. 2005, Eur J Appl Physiol 95(5–6):529–536; PMID 16151832; DOI 10.1007/s00421-005-0039-y).
• Demands are strongly shaped by stage/race type and by result — flat stages load short-duration maximal mean power, mountain stages load sustained efforts; higher race category and better placings carry distinct intensity signatures (Sanders & van Erp 2021 review, Int J Sports Physiol Perform 16(1):3–12; PMID 33271501; DOI 10.1123/ijspp.2020-0508; van Erp & Sanders 2021, 2,920 files / 20 pros, Eur J Sport Sci 21(5):666–677; PMID 32584197; DOI 10.1080/17461391.2020.1788651).

Training implication: average power under-describes a road race. Train the full record power profile — repeated 10 s–5 min supra-threshold efforts off fatigue — not just steady threshold.

Drafting changes everything

Sitting in the bunch is not optional physics — it's the sport. In a tightly packed peloton, drag on riders mid-to-rear drops to roughly 5–10% of an isolated rider's (Blocken et al. 2018, J Wind Eng Ind Aerodyn 179:319–337; DOI 10.1016/j.jweia.2018.06.011; no PMID — cite by DOI).

Implications:

• Watts spent out of the draft (attacking, chasing, riding the front, wind-exposed) cost multiples of watts spent sheltered — positioning is a physiological lever, not just tactics.
• The record that matters is what you can produce after hours of surge-and-soft-pedal, not fresh.

The finish: sprinting after a hard race

Most road races and criteriums are decided by a short, very high-power finish thrown after the race has already emptied the tank.

• Professional road sprints reached peak power of ~17.4 ± 1.7 W/kg, and power/cadence/speed rose significantly in the final minutes before the sprint — the sprint is launched off a high-intensity lead-in, not from rest (Menaspà et al. 2015, Int J Sports Med 36(13):1058–1062; PMID 26252551; DOI 10.1055/s-0035-1554697; sprinter performance analysis: Menaspà, Abbiss & Martin 2013, Int J Sports Physiol Perform 8(3):336–340; PMID 23038704; DOI 10.1123/ijspp.8.3.336).
• Prior stochastic high-intensity cycling impairs the finishing sprint: ~1 h of variable cycling cut subsequent 30 s sprint peak and mean power by ~5–6% vs constant-power riding (Etxebarria et al. 2019, Front Physiol 10:100; PMID 30837886; DOI 10.3389/fphys.2019.00100) — directly relevant to criteriums and to any bunch finish.

Training implication: train sprints in a pre-fatigued state, not just fresh — the race-day sprint is a fatigued sprint. This pairs with the fatigued-state work in durability.md.

Durability decides who is there at the end

Maintaining maximal power after high accumulated work is a key determinant of success in pro racing: category-1 climbers and sprinters showed smaller declines in discipline-specific maximal mean power after heavy accumulated work than lesser riders (van Erp, Sanders & Lamberts 2021, Med Sci Sports Exerc 53(9):1903–1910; PMID 33731651; DOI 10.1249/MSS.0000000000002656). This is the same durability quality profiled in durability.md (Muriel 2022; Spragg 2023) — in road racing it manifests as still having a sprint or a matching-move after 4 hours.

Criterium vs road race

Both are intermittent, but criteriums are shorter, more relentlessly repetitive (constant corner-surge-brake cycles, little true recovery), while road races add sustained climbs and longer selective efforts. The peer-reviewed criterium-specific literature is thin — the strongest indexed evidence is the intermittent-effort → impaired-sprint mechanism above (Etxebarria 2019); much criterium-vs-road detail lives in reviews (Sanders & van Erp 2021) rather than dedicated studies.

⚠️ Gap flag: treat criterium-specific numbers as under-documented; reason from the intermittent-effort principles here and verify against the athlete's own race files via MCP.

HR is unreliable here — power governs

Like MTB (mtb-xco-demands.md), road racing is intermittent: HR lags the surges that decide the race and drifts over hours. Prescribe and read intensity from power; use HR only for steady sections and recovery trends, screened for the confounds in data-confounds.md.

Practical emphasis (reason here, pull data from MCP)

• Repeatability of supra-threshold efforts — train repeated 30 s–5 min efforts off fatigue
• Fatigued sprint — practice the finish pre-tired, not just fresh (Etxebarria 2019)
• Durability — hold power after accumulated kJ (durability.md; van Erp 2021)
• Positioning treated as an energy-saving skill (Blocken 2018), not an afterthought
• Judge intensity by power, not HR (data-confounds.md)