SYNTH-41 · Guide
Submaximal fatigue test (SFT) in the warm-up
How to run it, how to read it, and what it can decide
The decision this framework makes
Whether a repeated fixed-load effort in the warm-up says the athlete is adapting, is fatigued, or says nothing today — and what that licenses the plan to change.
Key Findings — what the evidence actually shows
The reference test clamps heart rate and reads power. The LSCT: 6 min at 60%, 6 min at 80%, 3 min at 90% of HRmax, then 60 s of recovery. Stages 2 and 3 are reliable (typical error 1.3-4.4%), and stage-3 power relates to time-trial performance at r = 0.91-0.94, HRR only at r = 0.55-0.68 (REF-752).
The protocol apps detect is Science to Sport's, and it is a different test from the LSCT. 3 min at about 110% of FTP, the wattage frozen for the whole season; heart rate averaged over the last minute; straight after, RPE (Borg 6-20, expected 14-16) and an estimate of how many more minutes the power could have been held, which its authors say is often more sensitive than RPE (their claim, not tested in this base); five wellness questions; fortnightly, often before a scheduled workout (Posthumus, sciencetosport.com, 2025 — a coaching protocol, not a peer-reviewed validation).Heart rate is not meant to plateau here: it is read as a tachometer at a fixed load, together with the subjective answers. No study in this base validates this 3-minute variant; the LSCT is its validated relative.
Adaptation also came with a faster HRR — it rose in the 7 improvers and fell in the 8 non-improvers after a HIT block (REF-754, C+, retrospective regrouping of the REF-753 cohort). HRR moves in the same direction for opposite reasons.
Using the test to time hard sessions helped in one small study: 48 ± 42 s vs 8 ± 45 s 40-km TT improvement over two weeks, same four HIT sessions, different timing (REF-753, n = 15).
The review conclusion is multivariate: read output, RPE and HRR together before altering training (REF-758).
1. How to run it
Freeze the load for the season — an absolute wattage (or pace), not a percentage of a threshold that will be updated. A test whose load moves with FTP cannot show a change in FTP.
Pick one protocol and never mix them. The fixed-power practitioner test: about 110% of FTP for 3 minutes, no heart-rate target. The LSCT: clamp heart rate at 60/80/90% of a MEASURED HRmax and read power (REF-3494 on why not an age formula).The reliability data behind the ±3 bpm noise band come from graded protocols ending at 85-93% of HRmax (REF-756, REF-759); whether a 3-minute hold at 110% of FTP ends in that band is not measured in this base (assumption), and it is not a target to aim for.
Same place in the same warm-up, before any hard work, same equipment, ideally indoors. Heat alone raises heart rate at a fixed load (SYNTH-38, SYNTH-39).
Record four things every time: heart rate (or power) over the last minute of the effort, RPE straight after on the same scale, the athlete's estimate of how many more minutes they could have held it, and — when the minute after is easy — the heart-rate drop over it.
Repeat about every 1-2 weeks, and establish a baseline over the first few tests before interpreting any single one (practice: the cadence, the frozen load and the four recorded items follow the coaching protocol; no study in this base sets them).
2. How to read it — the combinations
| Output at fixed load | RPE | HRR (60 s) | Most consistent reading |
|---|---|---|---|
| HR lower (or power higher on a HR clamp), beyond noise, over several tests | same or lower | same or faster | Adaptation — when recent load was normal (reasoning; the only data are REF-754, C+: HRR rose and stage-2 power tended to rise in 7 improvers, RPE not reported) |
| HR lower / power higher | higher | faster | Acute fatigue or functional overreaching — not fitness (REF-757, REF-755, REF-208) |
| Cannot reach target HR (HR clamp) | higher | — | Fatigue (REF-755) |
| HR higher at the same power | higher | slower | Check the ordinary causes first: heat, poor sleep, dehydration, early illness, a change in usual caffeine (practice — no study in this card describes this combination; heat, sleep and illness as ordinary causes: SYNTH-39, SYNTH-38) |
| Change within ~3 bpm or within 1.3-4.4% power | any | any | No change — noise (REF-752: LSCT power error; REF-756: heart-rate noise on a graded running protocol over four consecutive days — the transfer to a fixed-power cycling test repeated weeks apart is assumed, and the band there may be wider) |
3. What it can decide
Go / not today for the key session — the one use with outcome evidence (REF-753: one B- study, n = 15, effect sizes only, on the heart-rate-clamped LSCT; its own timing rules are not in the abstract, so using the section-2 fatigue pattern as the trigger is this card's reasoning). On a fatigue pattern, swap today for easy riding and move the key session, keeping the block's total dose.
Whether a recovery phase has done its job — markers returned to baseline within ~3 days after an 8-day camp in six professional women (REF-755, C+; transfer to amateurs is assumed).
It does not set thresholds. A trend in the test is a reason to schedule a threshold test, not a replacement for one.
It has not been tested for trimming interval targets within the session. Lowering or raising today's interval watts by a few percent from the warm-up test is a plausible extension, not a finding; if done, say so and keep the change small.
Practical Application
Use a fatigue pattern to move the key session, not to cancel the block: easy today, the hard session on a later day, same total dose (REF-753).
Freeze the test load in absolute watts for the season — the common protocol is about 110% of FTP for 3 minutes — and ask RPE and "how many more minutes could you have held it" straight after. Treat a heart-rate change within about 3 bpm as noise (REF-756, measured on a graded running protocol over consecutive days — a transfer to this test).
Before interpreting any shift, ask about heat, sleep, illness, caffeine and the previous two days' load — they also move heart rate at a fixed load (SYNTH-39, SYNTH-38).
Act on a pattern across two or more tests, or on one test that agrees with how the athlete feels — not on one number (the base's own decision rule, reasoned from the studies above; not tested).
What NOT to do
Do not compare tests done at different loads, in different conditions, or outdoors on a hot day against indoor tests.
Do not rescale the athlete's threshold or zones from the test.
Do not present the 3-minute fixed-power variant as validated — the validated protocol is the LSCT.
How this entry was checked and changed
Why this level
(synthesis; the review of overreaching markers REF-207 is A, REF-758 and REF-208 are B+, the test studies themselves are B to C+, small and mostly from one research group; the only outcome study REF-753 is B-, n = 15; the fixed-power variant athletes run has no study; the decision rule for acting on a test is reasoning over them)
Content correction 2026-09-27
Resync 2026-09-28: Cross-reference "REF-361 (HRV-guided training — timing hard days from a morning measurement)" -> REF-361 is Plews 2013, a narrative review of HRV monitoring in elites; the entry itself points HRV-guided training to REF-362.
Verification 2026-10-01: 25 ref-bearing sentences/cells read against REF-752…759, REF-207, REF-208, REF-3494 and the pointed lines of SYNTH-38/39; 14 OK.Every number quoted from an entry matches it (LSCT 6/6/3 min at 60/80/90% HRmax + 60 s; typical error 1.3-4.4%; r 0.91-0.94 vs 0.55-0.68; ~3 bpm at 85-90% HRmax; HRR error 6-7 → 3 bpm at 86-93%; 48 ± 42 vs 8 ± 45 s, n=15; ~3 days after an 8-day camp). No NUMBER or WRONG finding.The core warning — lower HR and faster HRR with higher RPE is fatigue — is well supported (REF-207 A, REF-208 B+, REF-757, REF-755).
Verification 2026-10-01: rows counted = 5 reading-table rows + 4 'what it can decide' bullets. Evidence: fatigue row, cannot-reach-HR row, recovery-within-3-days. Mixed: adaptation row (C+, RPE and HR-at-power not in the entry), noise row (right numbers, different test/sport/spacing), go/not-today (outcome study exists, but its trigger rule is unknown). Practice: 'HR higher' row, 'does not set thresholds', trimming targets (this last one is correctly labelled).
Verification 2026-10-01: the structural weakness is that the test the athlete actually runs (3 min at ~110% FTP, fixed power) has no entry and no study in the base — the card says so twice — while all reliability, noise and outcome numbers come from heart-rate-clamped or graded protocols (LSCT cycling, HIMS/LSRT running). The card needs transfer clauses where it reuses those numbers.
Verification 2026-10-01: Tier line in the live body still reads 'Tier 2-4' although the 2026-09-28 log records a change to 'Tier 1-5' — the resync did not land in the snapshot.
Verification 2026-10-01: 'no caffeine where usual' as a cause of a HIGHER heart rate is physiologically odd and has no source; reworded neutrally rather than supported.
Verification 2026-10-01: no contradiction found with SYNTH-38 or SYNTH-39 (both agree heat moves heart rate at a fixed load).