KILYO
Skip to content

Race time predictor: can you trust your watch?

Your watch shows "half: 1:45" and you wonder whether to believe it. That number is neither a promise nor a fluke. Here is where it comes from, how far it holds, and how to read it.

The KILYO team

The sports coach that reads your workouts

6 min read
Trail runners seen from behind on a dirt path at the edge of a forest, in broad daylight, wearing hydration vests and race bibs.
Photo: Markus Spiske / Unsplash

The short answer

Treat your race time predictor as a guide, not a goal. A 2025 study of 154 amateur runners wearing a Huawei Watch GT Runner (Frontiers in Sports and Active Living) found an error under 3% from 5 km (3.1 mi) to the half marathon. The marathon was not tested, and individual gaps reached several minutes. The marathon is the most fragile distance, so leave yourself a margin.

Key takeaways

  • On the Forerunner 955, the manual says the prediction relies on your estimated VO2max and your training history from the last few weeks.
  • From 5 km (3.1 mi) to the half marathon, predictions are fairly good on average. For the half marathon, 95% of runners in the study landed roughly 4 to 9 minutes away from the prediction, on one side or the other.
  • The marathon is more fragile: Riegel's formula predicts correctly up to the half marathon, but poorly for the marathon.
  • Heat, altitude, and a new watch can shift the number without your fitness changing.

You just got back from a 10K, and your watch says you could run a half marathon in 1:45 and a marathon in 3:40. You want to believe it. You also have a question: what is that number based on, and what happens if you trust it on race day?

Good news: the prediction isn't random. It has real limits, and those limits are known. Once you know where the number comes from and when it goes wrong, you can use it for what it's worth: a guide.

What your watch actually shows

A race time prediction is an estimate of what you could do today, based on the fitness your watch thinks you have. It is not a guaranteed finish time.

The Forerunner 955 manual says it plainly: at first, predictions may be less accurate, because the watch needs a few races to learn your performance. With a new watch, the number mostly tells you what the watch doesn't know about you yet.

Where the number comes from

On the Forerunner 955, the manual says the prediction is based on two things: your estimated VO2max and your training history. The watch analyzes your data over several weeks to adjust its estimates. The manual does not detail the exact formula.

Another approach has been around for a long time: start from a shorter race time to estimate a longer distance. The best known is Riegel's formula: as the distance doubles, your pace slows, and the formula sets by how much. It's simple, but it knows nothing about your training or the conditions on race day.

What the studies show

A 2025 study published in Frontiers in Sports and Active Living followed 154 amateur runners wearing a Huawei Watch GT Runner for at least six weeks before testing. Predicted and actual times were very close for 5 km (3.1 mi), 10 km (6.2 mi), and the half marathon.

But "on average" doesn't mean "for you." Here is the gap between prediction and reality in that study. Bias is the group's average gap. The margin is the individual gap that covers 95% of runners.

DistanceAverage biasIndividual margin
5 km (3.1 mi)−20.4 s± 44.2 s
10 km (6.2 mi)4.1 s± 299.1 s
Half marathon143.8 s± 400.4 s

For the half marathon, 95% of runners therefore land roughly 4 to 9 minutes away from the prediction, on one side or the other, with an average bias of 2 min 24 s. Focus on the width of that range more than its direction. Over 5 km (3.1 mi), the margin is much narrower.

For the marathon, Vickers and Vertosick's 2016 study of recreational endurance runners is clear: Riegel's formula predicted times correctly up to the half marathon, but poorly for the marathon. Two models based on one or two earlier races did better, with a root mean square error of 228 and 208, compared with 381 for Riegel. The lower the number, the better the prediction.

Why the prediction goes wrong

Three causes come up again and again.

Your VO2max estimate may be off. If the starting estimate is wrong, the prediction built on it will be too. A validation of the Apple Watch on 30 participants, tested in a lab on a treadmill, found an average underestimate of 6.07 mL/kg/min, with large differences from one person to another. That study is about Apple, not Garmin, but an off VO2max shifts any prediction that depends on it.

Weather and location. On Garmin fēnix 7 watches, the manual says the VO2max estimate and training status get adjusted when the temperature goes above 22 °C (72 °F) and when altitude goes above 800 m (2,625 ft). A number that moves in midsummer or during a mountain trip doesn't necessarily mean your fitness has changed.

Too few reference races. With a new watch, it needs a few runs to learn your performance, according to the Garmin manual.

How to read the prediction without getting caught out

Studies measure how accurate predictions are. The advice below comes from practice, not a controlled trial.

Read the prediction as a range, not a single point. If your watch says 1:45 for the half marathon, think of a deliberately cautious range, say 1:35 to 1:55, and look at what could push you one way or the other: weather, course, how you feel that day.

Then cross-check it with real data. A recent race or a timed effort over 5 or 10 km (3.1 or 6.2 mi) is worth more than any estimate. To go from a short distance to a long one, Riegel's formula holds up to the half marathon and falls apart at the marathon.

Finally, watch the trend more than the number. A predicted time that improves week after week means something. A single number after a hot trip may mostly reflect the heat.

To see what a target means in pace and split times, try the pace calculator.

Free calculatorPace calculatorOpen the calculator

What you can do at your next workout

  • Note today's predicted time for your target distance, with the date.
  • If you feel good, add a 20 to 30 minute stretch at a harder effort after a warm-up, at an intensity you know you can hold. Skip it if you have pain, unusual fatigue, or are returning from a recent break.
  • Check again in three or four weeks: has the number moved, and why?
  • If a race is coming up, do a timed 5 or 10 km (3.1 or 6.2 mi) effort when you're fresh and well recovered. The goal is to measure your fitness, not to push the watch number up. If you have pain, talk to a doctor about it.

A prediction only makes sense with the context around it. KILYO reads your watch workouts, compares your pace, heart rate, training volume, and consistency, and adjusts your plan accordingly. If your watch promises a fast half marathon but your long runs are rare, the gap shows up in your workouts, and the plan takes it into account. You can also review how to prepare for a first half marathon or where to start for a first marathon.

Frequently asked questions

How long before a prediction is reliable with a new watch?
Garmin says it takes a few races for the watch to learn your performance. In the 2025 Huawei study, runners wore their watch at least six weeks before testing. So plan on several weeks of varied runs.
Why does my watch predict a slower time than my personal best?
It is based on your estimated VO2max and your last few weeks of training. Heat and altitude can also shift the number. Compare it with your last timed race rather than trusting the display alone.
Garmin, COROS, Apple Watch: which predicts best?
The only study cited here that measures prediction accuracy is on the Huawei, from 5 km to the half marathon. Each brand uses its own data, and none of the studies cited here compare them, so there is no way to rank them.
Should I aim for the predicted time on race day?
Aim for a range, starting at a pace you know you can hold. The Huawei study includes no marathon. For Riegel's formula, the Vickers and Vertosick study shows poor prediction at the marathon, so leave yourself a margin.

Sources

  1. Garmin (n.d.). Forerunner 955, user manual: viewing your race time predictions. Official page
  2. Garmin (2025). fēnix 7 Owner's Manual, Heat and Altitude Performance Acclimation. Official page
  3. Dai J., Chen G., Gu Z., Qi Y., Xu K. (2025). Accuracy of smartwatches in predicting distance running performance. Frontiers in Sports and Active Living. DOI
  4. Vickers A. J., Vertosick E. A. (2016). An empirical study of race times in recreational endurance runners. BMC Sports Science, Medicine and Rehabilitation. DOI
  5. Investigating the accuracy of Apple Watch VO2 max measurements: a validation study (2025). PubMed Central. Article

Written with the help of AI.

What your workouts say about you.

Your first review

  • Your training today
  • What stands out in your workouts
  • A next step to discuss

How detailed the review is depends on your history and the available data. Paid coaching comes after, if you want it.

Free to start · no credit card needed