REF-1280 · Haugen et al., 2018 · International Journal of Sports Physiology and Performance · Paper · PubMed

How much more can I improve over the years?

World top-100 track athletes peaked at 25-27 (marathon ~28-29); throwers gained 7.0% in the 5 years before peak, runners and jumpers less. Elite only; no amateur data (B+, from the abstract).

Main caveat: Read from the abstract only. Event-level results, per-athlete season counts and the modelling details could not be checked.
moderate evidenceMeasured on elite to world-class athletes

Haugen, Hopkins and colleagues quantified peak age and the improvement over the years before it in elite athletics, by performance level, sex and discipline.

They downloaded the individual season bests of world-ranked top-100 athletes from 2002 to 2016 from the International Association of Athletics Federations' website (14,937 athletes, 57,049 results) and fitted a quadratic curve to each athlete's performance-age data with a linear modelling procedure.

Mean peak age was typically 25-27 y, somewhat higher for the marathon and for male throwers (about 28-29 y). Women peaked later than men in the hurdles and in middle- and long-distance running (mean difference 0.6 ± 0.3 to 1.9 ± 0.3 y, ±90% confidence limits; small to moderate), and male throwers later than female throwers (1.3 ± 0.3 y; small).

Throwers showed the greatest improvement over the 5 years before peak age (mean 7.0%, SD 2.9%), "clearly ahead of" jumpers, long-distance runners, hurdlers, middle-distance runners and sprinters, for whom the abstract gives "3.4, ±0.2% to 5.2, ±0.2%; moderate to large".

Top-10 athletes improved more than those ranked 11-100 in every event group except throws (1.0 ± 0.9% to 1.8 ± 1.1%; small), and women improved more than men in every event group except sprints (0.4 ± 0.2% to 2.9 ± 0.4%).

Our reading: near the top of a sport, five more years of training buy a few per cent; the exact figure for runners is uncertain from the abstract alone, because the "3.4 to 5.2%" bracket can be read either as those groups' own improvements or as the throwers' lead over them (see Key Results).

Protocol

Data: individual season bests of world-ranked top-100 athletes, 2002 to 2016, from the IAAF website; 14,937 athletes and 57,049 individual results.
Analysis: a quadratic curve fitted separately to each athlete's performance and age data with a linear modelling procedure; outcomes were peak age and the percentage improvement over the 5 years before peak age; comparisons by performance level (top 10 vs 11-100), sex and discipline (sprints, hurdles, middle distance, long distance, marathon, jumps, throws); effects with ±90% confidence limits and magnitude labels.
Not in the abstract: the number of ranked seasons per athlete, minimum seasons for a curve, event-level values, how athletes with fewer seasons before their peak were handled, doping exclusions.
Funding and interests: not stated in the abstract or the PubMed record.

Key Results

Outcome (abstract)Result (mean, ±90% confidence limits where given)
Peak agetypically 25-27 y; marathon and male throwers about 28-29 y
Women vs men, peak age (hurdles, middle and long distance)women later by 0.6 ± 0.3 to 1.9 ± 0.3 y (small to moderate)
Male vs female throwers, peak agemen later by 1.3 ± 0.3 y (small)
Improvement over the 5 years before peak: throwers7.0% (SD 2.9%), the greatest of all event groups
Improvement over the 5 years before peak: jumpers, long-distance runners, hurdlers, middle-distance runners, sprinters"3.4, ±0.2% to 5.2, ±0.2%; moderate to large" — either these groups' own improvements or the throwers' lead over them (see below)
Top 10 vs 11-100, improvement+1.0 ± 0.9% to +1.8 ± 1.1% (small), all event groups except throws
Women vs men, improvement+0.4 ± 0.2% to +2.9 ± 0.4%, all event groups except sprints
An ambiguous bracket. Every other "±" figure in the abstract is a between-group difference (most with a magnitude label), while the throwers' 7.0% is given as a mean (SD).Read the same way, "3.4 to 5.2%, moderate to large" is the throwers' lead, which would put the other groups' own 5-year improvements at about 1.8-3.6% (7.0 minus 5.2 and minus 3.4, our arithmetic); read as the groups' own values, they improved 3.4-5.2%. Which group had which value is not in the abstract either. The full text would settle it; it was not read.

Mechanisms

Measured: world-ranking season bests over 15 years; per-athlete fitted performance-age curves.
Authors' interpretation: the peak ages and improvement rates by performance level, sex and discipline are useful for setting goals and judging strategies in elite athletics.
Inference: improvement shrinks as an athlete nears their ceiling, so the last five years before a world-class peak add only a few per cent; the top 10's slightly larger improvement fits continued development rather than an earlier plateau among the very best.
Inference: an athlete enters these lists only when ranked, so a career that stalled early or ended in injury contributes few seasons; the curves describe successful careers, and a quadratic forces a symmetric rise and fall around the peak.
Inference: the later peak in the marathon fits endurance qualities that keep developing with training years (consistent with REF-988 and REF-1003 in elite endurance athletes), but the paper does not test why.

Practical Application

How much more can I improve over the years? World top-100 track athletes peaked at 25-27 (marathon ~28-29); throwers gained 7.0% in the 5 years before peak, runners and jumpers less. Elite only; no amateur data (B+, from the abstract).
Near the ceiling, years buy per cents. For a highly trained athlete, a forecast of more than a few per cent over several seasons needs a reason; for professionals' season-to-season change see REF-1279 (critical power about 3%) (inference, transfer from world-class athletes).
Amateurs are not on this curve. An adult who started late and trains part-time is further from their ceiling and may improve faster and peak later; no study in the base measures it, so forecast from the athlete's own trend across blocks, not from these figures (not from this paper; the base's own practice, SYNTH-42: let the measured outcome decide).
Women's longer development: in these elite data women improved more over the 5 years before peak and, in hurdles and middle and long distance, peaked later than men; whether this holds for amateurs is not known.
Masters athletes: the abstract reports no figures for the decline after the peak; see SYNTH-35.

Limitations

Read from the abstract only. Event-level results, per-athlete season counts and the modelling details could not be checked; the bracket for the non-throwing event groups is ambiguous (see Key Results).
Survivorship. Only seasons in the world top 100 are recorded, so the curves describe careers while they were successful.
Fitted curves. Peak age and the 5-year improvement come from per-athlete quadratics, which force a symmetric shape around the peak.
World-class track and field only. Amateurs, cyclists and triathletes are not covered.
Funding and interests: not stated in the abstract or the PubMed record. The data are public world-ranking lists of the athletics federation (IAAF); no federation funding is stated.
How this entry was checked and changed
Why this level

large observational cohort (n = 14937 athletes, 57,049 results); held at B+ because only seasons in the world top 100 are seen, so careers are observed while successful, and the peak age and the 5-year improvement are read from fitted curves; read from the abstract (closed access; no Europe PMC copy, and the Victoria University repository record listed by OpenAlex is not open access and its host did not respond)