A model study of optimal training reduction during pre-event taper in elite swimmers
Participants
n=8 elite swimmers; model parameters were derived from their training and performance data over two seasons.
Protocol
A non-linear (fitness-fatigue type) model was fitted to each swimmer; the fit between modelled and actual performances was statistically significant for every swimmer (r² = 0.56 ± 0.06; P < 0.01)
Computer simulations then searched for the step and progressive tapers that would maximize performance either (1) after regular training only or (2) after overload training — a 20% step increase in regular training for 28 days
Key Results
Step taper after overload beat step taper after regular training: 101.4% (s = 1.6) vs 101.1% (s = 1.4) of personal record (P < 0.01)
…but needed a longer taper: 22.4 days (s = 13.4) after overload vs 16.4 days (s = 10.3) without (P < 0.05); the standard deviations are of the same order as the means, i.e. the optimal duration differed a lot between swimmers
Progressive taper improved simulated performance only after the overload period (101.5%, s = 1.5; P < 0.001)
Within the model, the negative influence of training (fatigue) was completely removed by the optimal taper regardless of prior training; the positive influence (adaptation) increased only after overload
Authors' conclusion: the characteristics of an optimal training reduction depend on the training performed in the weeks before the taper
Practical Application
Simulations on a model fitted to 8 elite swimmers: the best step taper reached 101.4% of personal record after a 28-day, +20% overload block vs 101.1% without it, but took 22.4 vs 16.4 days — an overload block before the taper buys a little extra in the model, and it costs a longer taper.
The optimal taper length varied widely between the 8 modelled swimmers (s = 13.4 and 10.3 days around the means) — taper duration is individual and depends on the preceding load, so treat any fixed "N-day taper" as a starting point, not a rule.
Inference, not tested: a progressive (gradual) reduction paid off in the model only after overload; after ordinary training the step taper did as well.
Not from this paper: the empirical pooled optimum — a 2-week taper with volume cut 41-60% and intensity and frequency kept — is in REF-107 (Bosquet 2007 meta-analysis). Combine that group optimum with this paper's message that prior overload lengthens the taper.
Not from this paper; the base's own practice: watch how the athlete feels and performs during the taper (freshness, short efforts) and shorten or extend by a few days rather than executing a fixed schedule.
Limitations
Modelling study, not a meta-analysis (an earlier version of this entry carried the wrong title and type). Its outputs are simulations from a fitness-fatigue model, not pooled empirical results.
Model fit was significant but modest (r² = 0.56 ± 0.06): the model explains about half of the variance in actual performance, and the simulated optima inherit its assumptions.
n=8 elite swimmers; parameters were estimated from their own two seasons of data. Transfer to another athlete or sport requires re-fitting, which is rarely done in practice.
The simulated tapers were never applied to the swimmers, so the predicted 101.4-101.5% gains are model outputs, not observed performances.
Use alongside the empirical taper evidence (REF-107) rather than as a substitute for it.
How this entry was checked and changed
Why this level
regraded 2026-09-27 from A: simulations on a non-linear model fitted to n=8 elite swimmers; the simulated tapers were not tested empirically
Content correction 2026-09-27
Title "Meta-Analysis of Tapering: Individual Factors Influencing Supercompensation" -> the paper's actual title, "A model study of optimal training reduction during pre-event taper in elite swimmers".
"2-4% improvement", "7-14 days optimal", "±1-3% individual variation", "15-20% non-responders", "training history / race-day motivation / taper anxiety predict response" -> deleted; none of it is in the abstract. Replaced with the abstract's numbers: n=8, r² 0.56 ± 0.06, +20% × 28 d overload, 101.4 vs 101.1% of PR, 22.4 vs 16.4 days, progressive taper 101.5% after overload only.
Application block (7-10 day taper, −40-50% volume, HRV by day 7-10, 5-day taper for "delayed adapters", 9-day plan) -> deleted as house advice; the empirical optimum is pointed to REF-107.
Level A -> B (model simulations on n=8); Tier "mixed (review/meta)" -> Tier 4 (elite swimmers); Sport general -> swimming.