Sweet spot, tempo and sub-threshold work

It takes one quality slot, never an easy day, and it is called what it is: no study tests sweet spot by name. No study in the base, and none found in a search of PubMed and Europe PMC (2026-10-29/30), tests sweet spot by name; one narrative review outside the base uses the term, as another name for the zone between the thresholds.

What exists are trials of training between the thresholds, or set by intensity domain. The base's default threshold session is this kind of work under another name: 4 × 10 min progressing to 2 × 20 min just below threshold, about 6-7 of 10 (SYNTH-27, B-).

Count the week before choosing. The number of quality sessions comes from SYNTH-26's table (B-; every count practice): none mandatory for a beginner; on three sessions one for a runner and two for a cyclist with events up to about 2-3 h; on four to seven usually two, a third only as a 2-4-week block; two to three on eight or more.

Threshold work takes one of them, two only above ~8 h (SYNTH-27, practice). It takes a quality slot, never an easy one: heart-rate variability recovered as slowly after 30 min between the thresholds as after 6 × 3 min above the second threshold, about 30 min against 5-10 min after up to 2 h below the first (REF-161, B, 9 highly trained men).

On two quality sessions, one sub-threshold and one above threshold; on one, choose by what the log lacks. No trial tests this pairing.

Why it stays: adding work above the ceiling to training kept below it raised VO2peak by 2.5 mL/kg/min more (95% CI 1.4-3.6), with no effect specific to intensity shown on peak power or time trial, partly for lack of sample (REF-989, A-, 14 studies); across 115 trials intervals raised VO2max more than continuous moderate-to-vigorous training (g 0.39, about 1.30 mL/kg/min) with no significant difference for thresholds or performance, on wide intervals and with moderate and vigorous work pooled (REF-1445, A-, abstract, sedentary to elite).

That the continuous work can therefore carry threshold gains is our inference, not that paper's test.

With one slot: if the last 8-12 weeks hold no work above threshold and the goal rests on aerobic power, go above threshold first (SYNTH-42, C+, its starting bet); if they do, and threshold power or a long steady effort is the goal, sub-threshold work is a fair lever (SYNTH-42's threshold row; REF-137, B+; REF-421, B).

More easy riding instead? Neither answer is proven, and the two do not compete for the same slot. Easy hours fill what is left of the week (SYNTH-51, B-).

Tempo in place of easy hours has no trial in cyclists; the nearest arm, 4 × 16 min at 88% of peak heart rate and lactate 4.9 mmol/L twice a week, was not detectably better than low-intensity riding alone over 7 weeks (+5.6 vs +4.2%, a composite, REF-150, B+, trained recreational cyclists, about 6 h a week before the trial).

More easy hours are not shown to add much either in a short block: 240 extra minutes a week of moderate-domain riding on top of three interval days showed no detectable gain over 6 weeks, about 23 h more riding (REF-1446, B, 19 riders, abstract; underpowered by its authors' estimate, more than 100 per group, and not a test of tempo), and across 41 studies weekly and total volume were not associated with gains in trained cyclists (REF-1443, A-, abstract; a regression across groups and studies in riders with VO2max at or above 59, not one rider adding hours).

The per-hour argument does not separate them: high-intensity intervals gave about 1.7 times the mitochondrial gain per hour of continuous endurance training, but that continuous work pools moderate and heavy (REF-465, A, markers, untrained to trained). A week with no quality session at all needs one before it needs either (step 3).

Set the target from a measured threshold. The athlete's measured critical power, MLSS, lactate threshold or tested FTP sets it, and feel and drift overrule it (§ Setting it). Zones and targets come from the athlete's own threshold; no band is a law.

Run the session by [SYNTH-27]'s ladder. 4 × 10 min, then 3 × 15, then 2 × 20, with about 2 min easy between reps, at about 6-7 of 10 (practice; the nearest tested shape is 5-7 × 10-14 min at RPE 14-15 on the Borg 6-20 scale, in one-week blocks, REF-137, B+; REF-421, B). Rep length by what can be completed (SYNTH-80, B-).

Progress by time at intensity, then raise the target after a retest (§ Progressing it).

Read the block at its review date (3-6 weeks, a house default, SYNTH-42) by the outcome chosen before it started: power or pace at the durations the goal needs, the same key session compared rep for rep, or a planned retest.

Stop relying on it when a block that was done stops paying, when the log lacks work above threshold, or when the goal or the athlete's hours change (§ When to stop relying on it).

Runners: the same logic, by critical speed or a field threshold pace. A planned between-threshold plan was not detectably worse than polarized in recreational runners (REF-1444, B+), easy-heavy beat between-threshold-heavy in sub-elite runners (REF-094, B), and SYNTH-47's run-load rules apply on top (B). At three runs a week SYNTH-26's table gives one quality run.

Use the athlete's measured number and name its protocol (SYNTH-53, B): critical power or critical speed first, then lab MLSS or lactate threshold, then a tested FTP labelled FTP20 and treated as a ceiling for long reps, not a target (SYNTH-27, B-). Fatigue moves the boundaries (PHYS-07, well supported), and an old number misplaces the session.

Aim below the ceiling, not at it. "A few percent below critical power rather than at it" is REF-750's application, not a result. Two tested places, neither shown to be best: 83 ± 6% of critical power (75-94), halfway between gas-exchange threshold and critical power, lactate 4.12 ± 1.30 mmol/L, all finished (REF-1447, B, acute, low-fit); about 66% of power at VO2max, 85 ± 3% of HRmax, lactate 2.8 ± 1.1 mmol/L (REF-137, B+, VO2max 70.5).

With only an FTP, a percentage is a starting guess. Start low, about 85-90% of FTP (practice, never a law; REF-137's intensity comes out near 83-88% of FTP only by our arithmetic, assuming FTP at 75-80% of peak aerobic power, on the group mean of 66%; the ~75% of ramp peak in SYNTH-53 is practice, the 80 is ours and no entry validates either; so it sits at or below the bottom of the coaches' band). Then the checks decide.

The checks overrule the number. About 6-7 of 10 on the app's scale, "hard but manageable" (SYNTH-27; RPE 14-15 on the Borg 6-20 scale in REF-137, the conversion being the base's own). Full sentences no longer comfortable, short phrases still possible (practice; speech first became difficult at the ventilatory threshold in REF-929, B).

The same power holds across the reps and heart rate climbs no more than about 4-5 beats from first rep to last below about 25 °C (a practitioner's rule of thumb, no entry; SYNTH-27). Lactate only where it is measured, on the same meter (SYNTH-53; PHYS-06, established): about 3 mmol/L in REF-137, and the 2-4.5 mmol/L band is REF-301's (B+, Tier 5).

A fade means too much. A rep that fades or climbs past those checks is above this athlete's ceiling today: lower the target or shorten the reps (REF-750's application; SYNTH-80). In heat, drift says less (SYNTH-38).

Minutes first, target second. The ladder in SYNTH-27; raise the target only after a retest. The suggestive data: in REF-1451 (B-, uncontrolled, veteran men, at MLSS and not below it) time at the intensity rose 43% by our arithmetic, distance about 50% and speed about 5%. More minutes alone are not shown to pay: 64 min of work a session at lactate 4.9 mmol/L did not beat low-intensity riding (REF-150, B+).

Do not hold one load for long. A fixed work rate stopped raising VO2max after about 3 weeks and rose again when the work rate was raised (REF-1218, B, n = 9, uncontrolled, probably untrained, VO2max and not threshold). The direction fits progressing inside a block; the interval is not a threshold result.

Time at this intensity is finite. At MLSS 44 min before training in veteran men (REF-1451) and 37.7 min cycling against 34.4 running (REF-1432, B+); at FTP20 a median of 35 min in recreational riders to 51 in professionals (REF-747, B); at critical power 20-41 min (REF-750, B). A hold to exhaustion is a test and not a session; a long continuous rep is for an event that asks for it, set below FTP20 (SYNTH-27, SYNTH-80).

Shorter reps make it easier at the same watts (5-minute reps lowered lactate by 1-1.5 mmol/L, REF-1053, B+); if the reps are shortened and the power is not raised, the session has been made lighter (SYNTH-80).

One axis at a time, and written down (SYNTH-42; what the outcome must beat is in SYNTH-93).

Nothing tests sweet spot by name. The 14 papers added for this guide all say so in their entries (REF-1442 to REF-1455), and REF-137 and REF-421 call their work moderate-intensity intervals.

Sub-threshold blocks moved threshold power in well-trained cyclists. +4.0 vs −1.3% at 4 mmol/L against time-matched regular training, with no difference in training impulse and no significant difference in 15-min power, p = 0.14 (REF-137, B+, randomised crossover, 30 men); +4.5 vs +2.1% against a high-intensity block, fixed order and no control (REF-421, B). Both are blocks of 6 sessions in 7 days; nothing is known of the once-a-week spread.

Long reps spread twice a week did not beat easy riding. +5.6 vs +4.2% on a composite of VO2peak, power at VO2peak and power at 4 mmol/L (REF-150, B+, about 9 riders per arm, a null). The same trial's 4 × 8 min gained +11.4%, more than either (P < 0.02), at lactate 9.6 mmol/L, a hard session and not a threshold one.

Training at 80-90% of VO2max moved the threshold more than training at 60-70% in elite skiers; that the upper band sits near the threshold is our inference. +3.2 ± 0.9% vs unchanged, between groups p = 0.04 (REF-1450, B-, abstract, allocation not stated).

It has a real between-group test, and says nothing of amateurs or of athletes for whom easy volume has stopped working.

In middle-distance runners three sessions at 82-92% of HRmax inside 50 km a week against one inside 70 km showed no between-group difference; speed at the threshold went 14.6 to 15.2 and 15.3 to 15.7 km/h, significant only in the first, and the changes (0.7 vs 0.4 km/h) were never compared (REF-1448, B-, 19 in the tables and 26 in the text).

A heavier middle was not detectably worse in recreational runners on very few hours, and that is not a time saving found by the trial. 24.8 vs 29.9 h follows from weighting a between-threshold minute double (REF-1444, B+); the plan still held 12.5% in zone 3, and session RPE was not lower (65.4 vs 60.9, not tested in the text). The four losses against one (two exclusions for adherence and two personal reasons) carry no fatigue data, so they say nothing on cost.

Adding work above the ceiling raised VO2max more than training kept below it; a performance payoff is not shown. REF-989 (A-) and REF-1445 (A-), with the limits in § What each intensity builds. Neither tests tempo.

A percentage does not place the session. § Setting it.

Time at the intensity can grow while the speed moves little, in one uncontrolled group. REF-1451 (B-). This is training at MLSS, not below it, and nothing compares ways of progressing.

It is a hard day by the one measure the base has. About 30 min of delayed autonomic recovery after 30 min between the thresholds, as after 6 × 3 min above the second (REF-161, B).

Work above critical power costs more later power, 10-20% against typically under 5% below it, a narrative range in which "below" is mostly low-to-moderate work and, in its two controlled trials, critical power and 12-min power held (REF-1239, B+); in 14 professionals 5 × 8 min at 105-110% of critical power cut short power more than about 2000 kJ below 70% of it, again not tempo (REF-193, B).