Marathon Pace Predictor

Race pace: {{ resultsReady ? formatPace(computation.values.pace_sec_per_km, 'km') : '—' }} Evidence: {{ evidenceBadge }}

Mean weekly distance:

Enter the eight-week mean, not only the highest training week.

Mean training speed:

Use the training-block mean rather than interval pace or one unusually fast run.

Personal calibration:

The neutral default is 0%. Calibrate from repeated race evidence, not a preferred goal.

Pace plan

{{ row.label }}: {{ row.display }}

Prediction method:

All displayed pace, speed, checkpoint, chart, and export values consume this canonical result.

{{ row.label }}: {{ row.value }}; {{ row.note }}

Evidence guide

Use the number cautiously

Compare the prediction with long-run durability, recent race results, course profile, weather, fueling, and health. Slow down or stop when race-day symptoms require it.

Calibration applied

The baseline changed by {{ formatSignedDuration(computation.values.adjustment_delta_s) }}. Keep this only when repeated marathons support the same direction and magnitude.

The chart renderer is unavailable. The same checkpoints remain available in the splits table.

Checkpoint	Distance	Elapsed	Average pace	Copy
{{ row.label }}	{{ row.distance }}	{{ row.elapsed }}	{{ row.pace }}

A marathon prediction turns recent training into a planning estimate, not a promise. The useful question is whether the training block supports a finish and pace that can be tested against long runs, shorter races, and race-day conditions.

Weekly distance describes how much running the body has repeatedly absorbed. Mean training speed describes the overall pace of that work, including ordinary mileage rather than only fast intervals. Both should come from the same representative eight-week period; combining a peak mileage week with one unusually quick run creates a training profile that never existed.

Pace and finish time are two expressions of the same even-effort plan. A pace of five minutes per kilometre can be extended across 42.195 kilometres to produce a finish estimate, and checkpoint times can be placed along that line. Real courses rarely behave that evenly because hills, turns, congestion, wind, heat, fueling, and fatigue shift effort throughout the race.

Evidence that strengthens or weakens a marathon pace estimate
Evidence	Stronger use	Weak use
Weekly distance	Mean from the same eight-week block	Highest week or planned mileage
Training speed	Total distance divided by moving time	Interval pace or one fast session
Personal adjustment	Repeated marathon bias in the same direction	Changing the answer to match a desired goal
Race check	Recent long runs and race results agree	The formula is used without outside evidence

The underlying study equation was developed from a small cohort whose reported marathon finishes ranged from 167 to 216 minutes. Predictions outside that baseline range are extrapolations. Even predictions inside it can be wrong for an individual because the model does not include course profile, weather, illness, injury, fueling, taper quality, or body composition.

Use the result as an informational starting point for race planning. It does not diagnose readiness, guarantee a safe finish, or replace medical advice, coaching judgment, event instructions, or the need to slow down or stop when symptoms require it.

How to Use This Tool:

Use one coherent training block and leave personal calibration at zero until repeated race evidence supports a change.

Enter Mean weekly distance from the representative eight weeks before the race and choose kilometres or miles per week.
Enter Mean training speed from that same block and choose kilometres or miles per hour. Use overall moving time, not interval pace.
Leave Personal calibration at 0% for the published equation. Apply a value from −12% through 12% only when prior marathons show a repeatable bias; a positive value makes the finish slower.
Review the predicted finish, pace per kilometre and mile, and evidence badge before using the checkpoint splits. Correct any range message instead of planning from a missing result.

Interpreting Results:

Study-range finish means the unadjusted prediction is at least 167 minutes and at most 216 minutes, matching the reported finish range of the original cohort. Extrapolated finish means the baseline falls below 167 or above 216 minutes. The badge evaluates the published baseline before personal calibration.

The checkpoint table assumes one constant average pace from start to finish. Treat its times as reference marks, not compulsory splits. A recent race, long-run durability, course elevation, forecast, fueling plan, or health concern can justify a slower plan even when the formula and evidence badge look favorable.

Technical Details:

The Tanda eight-week model predicts marathon pace from weekly distance in kilometres and mean training pace in seconds per kilometre. Faster average training and greater weekly distance reduce the predicted race pace, but the distance contribution follows an exponential curve rather than improving at a constant rate.

Formula Core

The baseline pace is calculated first, then extended across the official marathon distance. Personal calibration multiplies the baseline finish time after the published equation.

P_{m} = 17.1 + 140 e^{- 0.0053 K} + 0.55 P

T_{0} = 42.195 P_{m}, T = T_{0} (1 + \frac{A}{100})

K is mean weekly distance in km/week, P is mean training pace in s/km, P_m is predicted baseline marathon pace in s/km, T₀ is baseline finish time in seconds, and A is personal calibration in percent. The final pace is T divided by 42.195 km.

Marathon predictor normalization and display rules
Quantity	Rule
Miles to kilometres	1 mi = 1.609344 km
Speed to pace	Training pace = 3,600 ÷ speed in km/h
Personal calibration	Inclusive range from −12% through 12%; 0% leaves the baseline unchanged
Checkpoint time	Final finish seconds × checkpoint distance ÷ 42.195 km
Display precision	Finish, checkpoint, and pace times round to the nearest second; race speed shows up to two decimals

Weekly distance and training speed each accept values from 5 through 250 and 5 through 25 in the selected units. Those are input bounds, not validated training recommendations. The evidence badge uses the unadjusted finish and inclusive comparisons: 167 ≤ baseline minutes ≤ 216.

Limitations and Safety:

Model error can come from both the equation and the training summary supplied to it.

The original cohort was small and covered a limited finish-time range, so the evidence badge is context rather than a confidence guarantee.
Even-effort splits do not account for elevation, wind, heat, crowding, aid stations, fueling, fatigue, or a planned negative or positive split.
Personal calibration can hide a poor baseline when it is chosen to reach a goal. Keep it only when repeated race results support the same direction and approximate size.
Training and result calculations stay in the browser; no remote prediction service is used. A local calculation does not make an unsafe pace medically appropriate.

Worked Examples:

Representative metric block

An eight-week mean of 80 km/week at 10.5 km/h gives a training pace of about 342.857 s/km. With calibration at 0%, the equation returns about 297.291 s/km, a displayed pace of 4:57/km and a predicted finish of 3:29:04. The baseline lies inside the 167 to 216 minute study range, but the checkpoint plan still assumes perfectly even effort.

References:

Prediction of Marathon Performance Time on the Basis of Training Indices, Giovanni Tanda, Journal of Human Sport and Exercise, 2011.
Competition and Technical Rules, World Athletics, effective January 1, 2026.