The week before a goal race, most runners do the same ritual: open the watch app and stare at the predicted time. And if you own more than one device — or you’ve ever compared with a friend on a different brand — you’ve seen the problem. Garmin offers one number. Coros offers another, minutes away. Apple’s VO₂max estimate implies a third. They can’t all be right. It’s fair to wonder whether any of them are.
The honest answer: they’re all reasonable estimates of slightly different things, built on different assumptions — and every one of them leaves out the same crucial variable. Understanding what’s actually inside these predictions is the difference between using them well and being misled at the start line.
Why the numbers disagree in the first place
Every race prediction starts from an estimate of your aerobic engine — usually VO₂max — and every brand estimates it differently. Garmin’s comes from Firstbeat’s algorithm, which reads your heart rate against your pace on outdoor runs: if you hold fast paces at low heart rates, the estimate climbs. Apple takes a different route, estimating from workout data against a population reference chart — and its number is noticeably influenced by which activities you do, often reading lower for cyclists and mixed-sport athletes than for pure runners. Coros derives its own aerobic fitness score and converts it to times with a model it doesn’t fully document.
Three brands, three rulers. It’s common to see the same body carry a 49 on one wrist and a 40 on the other — runners on Apple’s own support forums compare gaps that large routinely. Neither device is malfunctioning. They’re answering slightly different questions with different data and different reference populations.
“Comparing your Garmin VO₂max to your Apple Watch VO₂max is comparing two different rulers. The useful question isn’t which number is right — it’s whether one consistent number is trending up or down.”
How accurate are they, really?
Reasonably — with caveats that matter. Validation research on wrist-based VO₂max puts the average error in the region of 2–3 ml/kg/min against laboratory testing, which is close enough to be genuinely useful for tracking fitness. But the error isn’t evenly distributed: studies have found it grows substantially for highly trained athletes — up to several times larger — and watch estimates tend to sit below lab values. Heart rate itself is noisy input: caffeine, stress, a poor night’s sleep, and wrist-sensor error all leak into the estimate.
The race-time conversion layered on top adds its own wobble. Early predictors simply mapped VO₂max to equivalent race times — times most recreational runners couldn’t hit at longer distances, because a strong engine doesn’t guarantee marathon-specific endurance. Modern predictors adjust for your actual training history, which helps, but the corrections can swing wide in both directions: side-by-side tests have caught one brand predicting a 10K more than two minutes faster than another on the same runner, the same week.
So the fair reading is this: your predictor is a good barometer of fitness direction, and a rough guide to potential — not a promise. Treated as a trend line, it earns its place. Treated as a pace plan, it will eventually burn you.
The variable none of them price in
Here’s the deeper problem, and it’s not an algorithm flaw — it’s a framing flaw. Every predicted time on every watch assumes a perfect day: cool air, low humidity, calm wind, flat course. The prediction is a laboratory number delivered to a world that isn’t one.
And conditions aren’t a rounding error. Once the dew point climbs into the high teens, the cost is real and compounding — a few seconds per kilometre at 5K scale becomes minutes at marathon scale. A runner whose watch promises 3:51 in clean air can run a perfectly executed 4:00 in warm humidity and conclude, wrongly, that they failed or the predictor lied. Neither is true. The number was honest; it just described a day that didn’t happen.
This is the piece to fix in your own head before race week: a prediction without a weather adjustment isn’t your race plan — it’s your ceiling. On a cool morning you may touch it. On a humid one, chasing it from the gun is the most reliable way to blow up at 30K.
How to actually use a race predictor
- Pick one source and stay with it. The absolute number matters less than the trend of a single, consistently-measured estimate. Cross-device comparisons only manufacture confusion.
- Read the trend, not the day. A prediction that improves across a training block is telling you something real. A number that jumped after one unusually cool tempo is telling you about the weather.
- Sanity-check against a race. A recent parkrun or 10K effort grounds the estimate in something your legs actually did. If the predictor and your race results disagree, believe the race.
- Adjust for the forecast — every time. Before you commit to a goal pace, ask what the dew point will be at the start. If it’s warm, the cool-day prediction is no longer your number, and pacing should start conservative.
- Respect the distance gap. Predictions get softer as the distance grows beyond what you’ve trained. A 5K forecast built on 5K training is solid; a marathon forecast built on 40km weeks is a hope.
The one-line version
Your watch’s prediction is your potential on a perfect day, measured by that watch’s particular ruler. Track its trend, verify it against real races, and reprice it for the weather before you trust it with your race.
Kovr prices your prediction in the weather you’ll actually run in.
Kovr builds its own VO₂max from your pace against heart rate on quality runs — one consistent number, seasonally adjusted, tracked the same way every time. Then it shows every distance twice: what you could run in cool air, and what today’s forecast will actually cost, from 5K to the marathon. The same engine as your heat cost, pointed at the start line.
Join the Kovr waitlistLaunching soon · Garmin, Apple Watch & Oura Ring
Frequently asked questions
How accurate is the Garmin race predictor?
Useful as a fitness barometer, not a race-day promise. Watch VO₂max estimates average within roughly 2–3 ml/kg/min of lab values, but the error grows for well-trained runners — and the predicted times assume ideal conditions and race-specific training.
Why is my VO₂max different on Apple Watch and Garmin?
Different rulers. Garmin reads heart rate against pace via Firstbeat; Apple estimates against a population chart and is influenced by which activities you do. Pick one source and track its trend rather than comparing absolutes.
Why did I miss my predicted race time?
Most often: race day wasn’t the perfect day the prediction assumed. Heat and humidity alone can cost minutes at half and marathon distance — a cost the watch never showed you. Training that lacked race-specific distance work is the other usual suspect.
Should I pace my race off the predictor?
Use it to set a realistic range, then reprice it for the forecast. If it’s warm or humid, the cool-weather number is your ceiling, not your plan — start conservative and let the day tell you what’s available.