SYNTH-10 · Guide
Pacing an event longer than ~2 h
Start below the day's average, set the cap from thresholds that will have slid by hour three, follow hills and wind with small swings, pace late hours by power or pace and effort rather than heart rate, and lower the target for heat; every percentage here is practice
The question it answers
How to pace an event over ~2 h: how to set the start, what drifts, how to ride hills and wind, when to back off, how heat and fuel change the target. Event builds are elsewhere: gran fondo and gravel SYNTH-50, long time trial SYNTH-62, ultra-cycling SYNTH-63, marathon SYNTH-64, road ultra SYNTH-65, mountain ultra SYNTH-61. Durability and its field test: SYNTH-32. Taper: SYNTH-34.
Default (what to prescribe first)
1
Start below the day's average. In events longer than 2 min, times may improve with more even pacing; beyond 4 h the optimum is not established and well-trained athletes typically pace positively (REF-010, B+, 2008 narrative review).
In 6589 gran fondo finishers the fastest group (under 6:00 h) had the lowest pacing range and started more conservatively (REF-899, B+, split times of one edition, association only). Not from a study; the base's own practice: ride the first fifth at about 90-95% of the planned average and stay under threshold whatever the group does.
2
Set the steady cap below fresh thresholds. After 2 h at 90% of VT1 power, power at the moderate-to-heavy transition was about 10% lower (217 → 196 W) while heart rate there was 9-10 bpm higher (REF-179, B, n=14 trained cyclists/triathletes). The fall accelerates: −5% at 139 ± 78 min on average (REF-143, B, n=12). Neither study gives a percentage to prescribe.
Not from a study; the base's own practice: past ~4 h, ride steady terrain under fresh VT1 power; in a 3-6 h event cap long climbs at or just under fresh threshold early and lower that cap as the hours pass.
3
Follow hills and wind with small, planned swings; spend nothing on unplanned surges.
4
Pace by power or pace plus perceived effort; heart rate is a cross-check (SYNTH-38, B).
Decision (who → what)
| Situation | Do this | Source and how sure |
|---|---|---|
| The first hour, mass start | Decide the opening pace before the gun; let faster groups go | Left to themselves, 7 cyclists opened a 16.1-km TT 14% above their mean power and faded (REF-849, B). In 501 runners over a 24-h race, better performers started slower relative to their own average (r = −.58 for the first 2 h; r = .03 for the last 2 h), and pacing shifted between editions, which the authors read as herd behaviour (REF-778, B+, observational). Across 1.7 million recreational marathoners, fast starts and fast finishes both went with slower times, with no effect sizes in the abstract (REF-820, B+, observational; plan in SYNTH-64) |
| Rolling or windy course | A little more power uphill and into the wind, less downhill and with it | Varying power with gradient at the same mean was 2.9% faster over an undulating 4 km (REF-848, B+, 20 experienced cyclists, ~7 min). Against the self-paced trial, 5% more power into a simulated headwind saved ~12 s over 16.1 km at the same mean power, with the lowest RPE (REF-849; the self-paced trial always came first). Models put larger gains on hilly and windy courses of 10-40 km (REF-795, REF-847, C, no riders). Untested beyond these short efforts |
| How big a swing | Small. Not from a study; the base's own practice: about 5%, more only on long climbs | ±5% swings cost nothing measurable over 1 h at 78% of VO2max (REF-241, B, n=8). 15% swings every 2 min raised VO2 and lactate at the same mean power (heart rate: no main effect in the Results (corrected 2026-10-12)), without a higher whole-bout RPE (REF-252, B+, 15 elite cyclists, 28-min bouts) |
| Descents | Keep working at a controlled effort | At the Ironman World Championship, athletes closer to their goal times held faster relative speeds on downhill segments and steadier heart rate from climb to descent (REF-659, B, observational, sample size not in the abstract) |
| Surges and attacks | Ration them from the start; late ones cost more | W′ was 18% lower by 80 min and 23% by 2 h of heavy cycling, and 60 g/h carbohydrate did not restore it (REF-145, B+, n=16). In 14 professionals, 5 × 8 min at 105-110% of CP lowered 1-s and 15-s power more than ~2000 kJ of low-intensity riding did, and only it lowered 3-min power; 12-min power and CP held (REF-193, B). Half of W′ recovers slowly (time constant 388 s; REF-131, B, n=10). Sitting in is the cheap alternative: a shielded cyclist needed over 30% less power at racing speeds (REF-842, C+, 1979, order of magnitude only) |
| Long-course triathlon bike | Plan the first lap so the fade is small | Six well-trained men faded from 239 to 203 W across an Ironman bike leg (REF-786, B, one flat race). Plan: SYNTH-57 |
| Over ~6 h and ultras | Far below fresh thresholds (inference from heart rate; no entry measured thresholds in an ultra); even effort; fewer stops | In 10 elite men over a 27-h race, HR/HRmax fell from 0.86 to 0.66 and averaged 0.68 (REF-797, B, HR only). Winners of a 161-km mountain race varied speed less than other top-5 finishers (CV 12% vs 14-15%; REF-829, B). In a 106-km mountain ultra faster finishers stopped less (r = −.772; REF-664, B, n=15). Plans: SYNTH-63, SYNTH-65, SYNTH-61 |
What drifts, and what to read
Power or pace at a given effort falls. See Default 2. 6-min power was ~10% lower after 4 h of race-like riding at 100 g/h (REF-144, B, 12 elite men).
Heart rate stops describing the effort. At the transition it rose 9-10 bpm after 2 h (REF-179). In heat it drifted 11% in 30 min against 2% in the cool, with sweat losses replaced (REF-3496, B+, n=10 well-trained men). In ultras it falls (REF-797, REF-664); in the one study that recorded it, perceived exertion climbed at the same time (REF-664). Inference: a fresh heart-rate cap is too low after 2 h, unreliable in heat and meaningless late in an ultra.
Perceived effort rises for reasons other than the legs. Mental fatigue shortened time to exhaustion by about 15% with physiology largely unaffected (REF-370, B+, n=16). A review of 11 experiments found the same direction (REF-371, A-). The flush model proposes that ultra pace follows perceived exertion held below a security reserve; the paper does not test it (REF-777, B+, running).
Heat and fuel as pacing inputs
Heat. At 32 °C vs 19 °C, with road-speed airflow and 72 g/h glucose, a ~15-min TT after 2 h was 15% slower and exogenous glucose oxidation 14-15% lower (REF-461, B, n=10 trained men). Among top-5 finishers of a 161-km mountain race, speed variation rose with maximum temperature (REF-829). No entry gives a per-degree correction, so the base prescribes none.Not from a study; the base's own practice: pace by perceived effort with cool-weather power or pace as a ceiling, and cut the opening hour most. Pre-cooling helped prolonged exercise in hot conditions (d = 1.91; REF-111, A-). Acclimation: SYNTH-04.
Fuel. Carbohydrate holds the cap, not the reserve. 60 g/h negated the 9% fall in CP at 2 h (REF-145). After 150 min, VT1 power fell 3% with carbohydrate and 6% without, while a 5-min effort fell in both (REF-181, B, n=12). The base no longer gives one rate for everyone: SYNTH-02 sets grams by duration and level.
When to back off, and when to push
Not from a study; the base's own practice:
Back off when effort at the planned power or pace climbs earlier than rehearsed, when the cap cannot be held on steady terrain, or when eating stops. Ease for 10-20 min, eat, then reassess. Before halfway, resolve any doubt by easing.
When NOT to use it / red flags
Do not quote the lab declines as an athlete's own numbers. They come from single protocols in trained riders.
A fast start is not time in the bank. The fastest gran fondo riders started conservatively (REF-899).
Illness, chest pain or dizziness are not pacing problems. House rule, not from a study: with fever or its company (chills, aches, malaise), a chesty cough, breathlessness out of proportion or vomiting, do not start, and stop if it appears mid-event. Chest pain, palpitations or dizziness: stop and see a doctor. A head cold with no fever allows easy and short only, which a paced event is not — the return rules are in SYNTH-36.
What the evidence does and does not show
Shows (B/B+, lab and short field trials in cyclists): thresholds and W′ fall within 2 h, and carbohydrate protects the first but not the second (REF-179, REF-145, REF-181); modest power variation with gradient was faster than constant power (REF-848), and with wind faster than a self-paced first trial (REF-849).
Does NOT show: a tested pacing plan beyond 2 h; the share of threshold an amateur can hold for 3-8 h; a heat correction; power declines beyond ~4 h. Women are few throughout.
Evidence gaps
A pacing intervention in events over 2 h (even vs conservative-start vs self-paced), in any sport.
Sustainable fraction of fresh threshold by duration, with power, in amateurs and in women.
Heat-specific pacing: only the two-temperature contrast of REF-461.
Variable-power pacing over hours on real climbs. Field evidence stops at ~7 min (REF-848).
Long-course triathlon pacing and its effect on the run. REF-786 has no run data.
How this entry was checked and changed
Why this level
(race pacing rests on B+ observational datasets and one B+ narrative review; the drifts on small B/B+ laboratory studies; terrain and wind on one short B+ field trial, one B lab trial and C models; start percentages, caps and back-off rules are house practice)
Content correction 2026-09-27
"CP declines ~9% per 2h without CHO; maintained with ≥ 60-90 g/h (REF-145)" -> CP −9% at 2 h on placebo (unchanged at 40 and 80 min), negated by 60 g/h (the only dose tested) without sparing glycogen; W' −18% at 80 min and −23% at 2 h, untouched by CHO.
"Neuromuscular fatigue (REF-162)" -> REF-306 (REF-162 is its duplicate); claims restated per the corrected entry: same limiting milieu above CP, M-wave fell similarly in all domains, neural drive rose only above CP.
"Pacing theory (REF-373)" -> REF-010 (REF-373 is deprecated); REF-010 now says more even pacing >2 min, >4 h unknown, positive pacing typical in ultras — the 90-95% start, "even RPE" and Borg targets labelled practice (REF-373's tombstone names them coaching extrapolation).
"2-3% of benefit from concurrent strength training (REF-074)" -> REF-074 gives performance SMD 0.52 in runners; REF-071 gives TT and economy gains without a pooled magnitude; percentage withdrawn.
"HR drift up to +10-20 bpm", "hydration 400-800 mL/h", heat "−5-10% / −10-20%" -> labelled practice; hydration traced to REF-061 (500-800 mL/h with sodium); heat anchored to REF-461 (TT 15% slower at 32 °C).
Added REF-193 (intensity of prior work, not quantity), REF-143 (GET −18% by 2 h), REF-144 (glycogen −36% at 100 g/h), REF-142 (Ironman substrate budget), REF-131 (W' τ-slow 388 s) where the framework leaned on them without citing.
Durability protocols, taper schedule and the hour-by-hour cascade labelled practice / inference.
(sync 2026-09-27, after the full-base pass) REF-193 "power profile −8 to −16% after 2000 kJ HI vs −1 to −3% after 2000 kJ LI" -> the corrected entry: HI was significantly less work, lowered 1-s/15-s power vs LIC and 3-min power vs fresh, 12-min power and CP unchanged; no percentages. Restated (now B).
REF-141 "economy degraded 2.1% vs 4.7% after strength (Sunde)", "CP loss <1% to −32%", "90-120 g/h", "Z1 volume / HIT levers", "names the Type-I-first mechanism" -> none in the corrected entry (now B+); replaced by its three suggestions (consistency/volume, prolonged sessions with race-pace bouts, heavy strength + plyometrics) and "large, unquantified variability".
REF-142 "344-519 g CHO from ~400-600 g stores" and the "sharp drop" framing -> deleted from the entry; replaced by its level-matched strategy argument; the sharp-drop line is now inference only.
REF-061 "500-800 mL/h with sodium" -> the consensus says start euhydrated and minimize dehydration; the rate is base practice.
Level parenthetical "A-level primaries" -> the primaries now range A to B (REF-141/010/370 B+, REF-193/143 B); B+ kept for the framework.
Cross-references to duplicate stubs repointed to their canonical entries (REF-162->REF-306, REF-373->REF-010); the stubs were emptied on 2026-09-27.
Level B+ -> B: the parenthetical claimed A-level durability, pacing and fatigue primaries; those are B+/B.
Descriptions of REF-144 repeated its pre-2026-09-26 version (muscle biopsies, glycogen/fibre-type depletion, fat oxidation doubling, power "preserved/maintained" at 100 g/h). The corrected entry took no biopsies and reports 6-min power FALLING ~10% after 4 h at 100 g/h; descriptions updated.
Resync 2026-09-28: REF-143 "GET power fell to ~82% of baseline by 120 min" (repeated as "GET −18% by 2 h" in the middle-phase, 3-4 h and red-flag lines) -> the corrected entry has no such table: VT1 power fell non-linearly, −5% at 139 ± 78 min, task failure at 234 ± 66 min, in 12 trained cyclists (B). The earlier note in this section quoting "GET −18% by 2 h" is superseded.
Resync 2026-09-28: "REF-143's entry uses a >10% HR rise over 3 h as a house marker" -> removed from REF-143; the HR-drift figure stays base practice without that attribution.
Resync 2026-09-28: REF-370 "time to exhaustion cut by 12%, physiology unchanged" -> 754 → 640 s, about 15%; physiology "largely unaffected".
2026-10-01: rewritten in the decision-framework format as the base's pacing framework for events over ~2 h; the four-domain fatigue model and hour-by-hour cascade are replaced by five pacing rules.Resolved in favour of the newer frameworks: flat "90-120 g/h" and "confirm 100-120 g/h" removed, grams per hour point to SYNTH-02; "500-800 mL/h with sodium" removed, drinking points to SYNTH-68; the monthly "depleted-state" rides withdrawn (SYNTH-69, SYNTH-32, SYNTH-50); the third last-two-weeks list removed, taper points to SYNTH-34; durability test and training protocols left to SYNTH-32.Deleted as house interpolation with no entry behind them: heat bands "−5-10% / −10-20%", "+10-20 bpm" drift, Borg targets. The 90-95% start stays, labelled practice in its own sentence. Folded in: REF-899, REF-778, REF-820, REF-848, REF-849, REF-795, REF-847, REF-241, REF-252, REF-659, REF-786, REF-797, REF-829, REF-664, REF-777, REF-179, REF-181, REF-3496, REF-062, REF-842.
Verification 2026-10-01: 31 cited REF entries read in full in the snapshot; every number, n, design and level tag next to a refId matches the entry head block (no LEVEL, no WRONG, no SOURCE findings). REF-820 and REF-899 have a DOI but no PMID; REF-241 a PMID but no DOI — not flagged.
Verification 2026-10-01: pointers checked — SYNTH-38 is level B; SYNTH-62 scope is 50-120 min; SYNTH-50 carries the same 'cap long climbs at or just below fresh threshold' practice (so the 'at' wording should be fixed there too); SYNTH-63 says 'Below-VT1 pacing from the start' for 6-12 h, which is stricter than SYNTH-10's 'at or under fresh VT1' past ~4 h.
Verification 2026-10-01: Default items 3 and 5 carry no refId and no practice label in the sentence itself; their support is in the table and heat section, so not flagged as UNLABELLED.
Row split counts the 5 Default items, 7 decision rows and the back-off/push block.
2026-10-01: the illness line now carries the house rule in full (fever or its company = stop; above the neck and no fever = easy and short only; chest pain, palpitations, dizziness = a doctor), consistent with SYNTH-36 and SYNTH-30.