REF-1279 · Valenzuela et al., 2023 · International Journal of Sports Physiology and Performance · Paper · PubMed

My season-best power hardly changed — have I plateaued?

In 61 pros, season-best critical power varied ~3% (CV 3.3%) and ≤1-min bests ~6% season to season. A few % is normal; pros, not amateurs (B, from the abstract).

Main caveat: Read from the abstract only. Absolute powers, the critical-power model, file counts and the data source could not be checked.
limited evidenceMeasured on elite to world-class athletes

Valenzuela, Pallares and colleagues asked how much a professional cyclist's peak performance, read from the record power profile, varies between seasons. They analysed power data from training sessions and competitions of 61 male professional cyclists (age 26 (5) y) over a median of 4 consecutive seasons (range 2-12).

For each season they took the highest mean maximal power (MMP) for durations from 10 s to 30 min and the critical power derived from them, assessed within-cyclist variability between seasons, and set the upper threshold of expected change at twice the normal coefficient of variation (CV).

Season-best MMPs agreed well and varied little between seasons (ICC .76-.88; CV 3.2-5.9%), with the lowest variability for efforts longer than 1 min. Critical power had an ICC of .79 (95% CI .70-.85) and a CV of 3.3% (95% CI 3.0-3.7%). The upper thresholds of expected variation were below 12% for efforts up to 1 min and below 8% for longer efforts.

The authors conclude that expected variation is about 6% for short and about 3% for long efforts, and that changes larger than 12% and 8% respectively are infrequent.

Our reading: for a professional, a season-best critical power within about 3% of last year's is normal variation, not a breakthrough or a decline; the CV mixes real change in form with differences in how often the rider went maximal, and it says nothing direct about amateurs, whose season bests are less often true maximal efforts.

Protocol

Design: longitudinal observational; n = 61 male professional cyclists, age 26 (5) y; power data from training sessions and competitions; median 4 consecutive seasons (range 2-12).
Per season: the highest mean maximal power at durations from 10 s to 30 min (the record power profile) and the resulting critical power (model and durations not in the abstract).
Analysis: within-cyclist variability between seasons (ICC, CV); the upper threshold of expected changes set at twice the normal CV.
Not in the abstract: absolute power values (W or W/kg), team levels, power meters and calibration, the source of the files, the number of files per season, how seasons interrupted by injury were handled.
Funding and interests: not stated in the abstract or the PubMed record.

Key Results

Measure (abstract)ICC between seasonsCV between seasonsUpper threshold of expected change (2 × CV)
Season-best MMP, 10 s to 30 min.76-.883.2-5.9%, lowest for efforts > 1 min< 12% for efforts ≤ 1 min; < 8% for longer efforts
Critical power.79 (95% CI .70-.85)3.3% (95% CI 3.0-3.7%)(a long-effort measure; its own threshold not given separately)
Authors' summary: expected variation about 6% for short (≤ 1 min) and about 3% for long efforts; changes greater than 12% and 8% respectively are infrequent.

Mechanisms

Measured: season-best field power by duration, within each rider, across consecutive seasons.
Authors' interpretation: "real-world" peak performance shows low variability between seasons in male professionals, especially for long efforts, so the record power profile can be used to monitor them.
Inference: the between-season CV is not measurement error alone: it holds real change in form, the racing programme and the rider's role, injuries, and whether a maximal effort at each duration happened that season.
Inference: a change between two seasons has an SD of about 1.4 × CV (the square root of two, our arithmetic), so "twice the CV" is about 1.4 SDs of a change and, under a normal model, is crossed by roughly one season-to-season change in six (our arithmetic) — "infrequent", not "rare".
Inference: short efforts vary more because sprint and anaerobic bests depend more on race situations and short-term form.

Practical Application

My season-best power hardly changed — have I plateaued? In 61 pros, season-best critical power varied ~3% (CV 3.3%) and ≤1-min bests ~6% season to season. A few % is normal; pros, not amateurs (B, from the abstract).
For a rider near the ceiling, year-to-year gains are small. A season-best critical power within about 3% of last season's is normal variation in professionals; promising a highly trained rider much more than that in a season needs a reason (inference, transfer from professionals).
Amateur season bests are a noisier yardstick. They depend on how often the athlete went all-out at that duration; compare like with like — the same duration, from a race or a planned maximal test (not from this paper; the base's own practice, SYNTH-42: field bests from efforts that were not maximal do not count).
Use it as a noise band, with its source named. Not from this paper; the base's own practice (SYNTH-42): a change inside its noise band is no change. For laboratory tests the band is REF-1278 (0.9-2.0%); for professionals' field critical power from season to season, this paper (about 3%).

Limitations

Read from the abstract only. Absolute powers, the critical-power model, file counts and the data source could not be checked.
Men, professionals only. Women and amateurs are not covered; amateurs' season bests are probably more variable (inference).
Season bests are not tests. They depend on races, roles and effort opportunities that vary from season to season; the CV mixes real change with opportunity.
The "upper threshold" is twice the CV, which by our arithmetic is crossed by about one season-to-season change in six under a normal model, if the CV is the within-rider SD of one season's best (the abstract does not say how it was computed); it is a heuristic, not a 95% limit.
Funding and interests: not stated in the abstract or the PubMed record; where the power files came from (teams, a platform) is not stated in the abstract.
How this entry was checked and changed
Why this level

an observational cohort of 61 (below the rubric's 200 for B+), with real-world season bests rather than a standardised test, professionals only; read from the abstract (closed access; the Universidad Europea repository record listed by OpenAlex is restricted access)