How age grading is computed

Age grading converts a race time into a percentage of the world-best time for someone of your sex and age at that exact distance. The intent is to make a 30-year-old’s 5K and a 65-year-old’s marathon comparable on a single axis: how close this performance is to what is humanly possible at this age.

Two reference systems are widely used. RaceRecords applies them by surface:

• Track events use WMA 2023.
• Road events use Alan Jones 2025.

The version applied to a given result is shown as a chip next to the percentage on the race detail page (e.g. wma-2023 or aj-2025).

For each event we read two numbers from the table that matches the surface:

• Open-class standard: the world best time at your sex and the specific distance, with no age adjustment.
• Age factor: a multiplier that reflects how much an average athlete slows down at your exact age for that distance.

The percentage is computed in two steps:

age_graded_time = actual_time × age_factor
percentage      = open_class_standard / age_graded_time × 100

Higher is better. Above 100% would mean faster than the open-class world record adjusted for your age, which is rare.

A 70%+ score is generally considered local class; 80%+ is national class; 90%+ is world class. Below 18 the tables do not apply, so the AG% column is left blank for any race you ran before your 18th birthday.

Separate from the AG% rating above, RaceRecords also predicts what you could run at other distances — the cross-equivalent grid, the race-detail equivalents, and the projections on the target-to-beat page.

These use Riegel’s endurance formula applied to your own time, not age-grade equivalence. Holding your AG% constant across distances would scale your time by the ratio of the two open-class world bests — an elite endurance ratio that predicts the marathon far too fast for ordinary runners. Riegel models the slow-down that comes with distance:

predicted_time = your_time × (target_distance / source_distance) ^ 1.06

When the prediction is for a different age than the source race (for example an old PR shown at your current age in the grid), the result is age-adjusted along that distance’s factor curve. Predicting another distance at the same age is plain Riegel on your actual time.

Ranges. The cross-equivalent grid shows a single number at 1.06. The race-detail equivalents and the target-to-beat projections show a central estimate with a range flanking it, spanning exponents 1.05–1.08. The exponent assumes distance-specific endurance — a 10K time can’t see whether you’ve done the long runs a marathon needs, so treat marathon predictions as a target, not a promise. A ~ marks wide distance ratios (over 2×) where any model is least reliable.

Personalized exponent. The 1.06 exponent is a population average. Because Riegel inverts cleanly, a peak block — a PR or age-grade PR at a shorter long-road race (Half Marathon, 10 Mile, or 15K) and again at the Marathon within 12 months — reveals your own endurance-fade exponent directly: k = ln(t_marathon / t_short) / ln(d_marathon / d_short). We look back five years for that block and, when more than one qualifies, take the most recent — and within the same timeframe your strongest short race by age grade, so a softer effort doesn’t understate your fade. Your k then becomes the central estimate — the bold number on the race-detail equivalents and the seed dot on target-to-beat — for endurance-distance targets (15K and up); shorter targets stay on 1.06. The 1.05–1.08 range still flanks it, stretched to include k when k sits beyond it. To keep one bad day from skewing things, k is only used when it lands in the credible range 1.045–1.12; a value outside that (a soft short race, a mismeasured course, or a marathon blow-up) is ignored and the generic 1.06 is used instead.

Source: Peter Riegel, “Athletic Records and Human Endurance” (1981).

World Masters Athletics 2023 Age-Grading Tables. The reference applied by Masters track-and-field bodies worldwide. Used by RaceRecords for the five supported track distances: 1500m, mile, 3000m, 5000m, 10000m.

The 2023 revision updated the open-class world bests and the age-factor curves against more recent performance data. It supersedes the WMA 2015 and WMA 2010 tables that may still appear on older calculators.

Authoritative source: world-masters-athletics.com / age-grading tables.

Alan Jones 2025 Road Age-Grading Tables. Alan Jones has maintained road-specific age-grading tables for decades; the 2025 revision is the current reference for the eight supported road distances: 5K, 10K, 15K, 10 miles, half marathon, 25K, 30K, marathon.

Road tables differ from track tables because road performance has diverged from track performance at the sub-elite level since the mid-2000s. Using a single track-derived curve for road events would systematically underweight road-only specialists. Alan Jones’s table is the most widely used road-specific alternative.

Authoritative source: github.com / AlanLyttonJones / Age-Grade-Tables.

Uncertified courses. An age-graded percentage assumes the stated distance is accurate. If you tick course not officially measured when logging a race, the AG% is still computed but flagged as uncertified course. Track results are always treated as certified because the surface is a measured oval.

Age-band steps. WMA and AJ both use single-year factors, so the percentage moves smoothly with each birthday rather than stepping at the masters age-band boundaries (W35, W40, W45 and so on). Most percentage changes between two races come from the factor changing by year, not from band transitions.

Below age 18. The tables only apply from age 18. Junior races are stored and listed, but no percentage is calculated.

Versioning.RaceRecords always grades every race against the current published table, so all of your percentages stay on one comparable scale. When WMA or Alan Jones publishes a revision, we adopt it and every result — past and present — recomputes against the new table; the version chip on each race (wma-2024, aj-2026, etc.) updates to match. A revision typically moves a given percentage by under one point, usually slightly down, because newer tables fold in faster world bests and tougher age factors. We keep the underlying race time untouched — only the derived percentage moves.

The grading-table data and the algorithm are summarised on this page in full. If you spot a discrepancy between the published tables (WMA 2023, Alan Jones 2025) and what RaceRecords applies, email support@racerecords.run with the input, the expected percentage from your reference, and what RaceRecords returned. Acknowledgement target: 72 hours.