Field note
Read the answer as a planning range built from your own comparable workouts, not as a map app's expected travel time and not as a promise you will arrive on the minute.
Moving time is not wall-clock time. A pause at a crossing, a café stop, or a recording that never marks pauses can make elapsed time longer than the minutes you actually spent moving. The estimate is intent. Your completed workout still comes from what the device recorded.
Runs inform runs. Walks inform walks. Hikes inform hikes. Scale each comparable session toward the planned distance, then give more weight to workouts that are newer, closer in length, and cleaner on distance-and-time evidence. Overnight vitals are not pace inputs.
A weighted middle beats a loud average. Huber (1964) laid out why robust location estimates resist outliers that would drag a mean. Recreational race-time work by Vickers and Vertosick (2016) is a reminder that endurance times are noisy and person-specific. Show a range, not a single heroic digit. When only one comparable workout exists, staying quiet is more honest than inventing precision.
| Input | Honest use |
|---|---|
| Your recent same-sport workouts | Central estimate and range |
| Map provider travel time | Not a substitute for your pace history |
| Hand-drawn unmarked ground | Keep the line length; widen the range |
| HRV or recovery score | May shape today's effort call; does not multiply pace |
How you spread effort across a session is a different question; see What is pacing?
Sampling a resolved route and reading a terrain tile under each point estimates ascent and descent. That is modelled ground height, not the elevation your phone or barometer measured on the day. Minetti and colleagues (2002) measured how steep uphill and downhill change the energy cost of walking and running. Wood and colleagues (2023) improved hiking-speed prediction with data-driven terrain context. Grade-adjusted pace uses the same cost idea looking backward; see What is grade-adjusted pace?
Bridges, tunnels, cliffs, and recent earthworks can disagree with the model. Unverified hand-drawn crossings still contribute measured line length, but confidence should drop. The line does not reveal mud, gates, crowds, or whether the path is open today.
Recent history should dominate. A common field choice is to prefer the last couple of months of same-sport sessions, and only reach further back when the recent sample is still too thin. Those windows are calibration choices to backtest, not physiological laws. Older evidence, distant-length comparisons, or a wide spread should lower confidence even when a time still appears.
Kiviniemi and colleagues (2007) and Vesterinen and colleagues (2016) tested heart-rate-variability-guided endurance prescription over time. That is a different question from predicting how many seconds one route will take today. Recovery evidence may shape the recommended effort. It should not quietly multiply your route pace.
Heat and humidity still move heart rate at any pace; see Why does heat raise my heart rate? Conditions, choices, and the ground itself get the final say.
Treat the number as a planning range from your own history. Leave provider travel times for driving. Leave arrival promises for trains.
I write these because I build Aera, an iPhone app that reads Apple Health against your own baseline and puts the research and the error bars underneath every number it shows you, including this one. You do not need it to use anything above.