How Garmin and Strava Try to Predict Your Race
Garmin and Strava tackle race prediction accuracy from two very different directions. Garmin’s Race Predictor, available on many of its running watches, converts your estimated VO2 max into pace equivalents for common race distances. It layers in basics like age, gender, and recent training, but fundamentally assumes a textbook-perfect race: ideal weather, flawless pacing, full taper, and solid fueling. Even when Garmin flags heat or altitude on its VO2 max widget, those factors do not carry into the standard Race Predictor, meaning it is closer to an aerobic ceiling than a realistic race-day outcome. Strava’s Performance Predictions lean on AI and large-scale activity data instead of VO2 max. Its model analyzes over 100 attributes across your own training history and similar runners, recalculating after each upload—provided you have at least 20 runs in a rolling 24-week window. Each distance is predicted independently, though it assumes a flat, track-like course and ignores terrain or altitude.
The Brooklyn Half: When Predictions Meet Reality
To see how these running watch predictions hold up, the author ran the Brooklyn Half Marathon wearing a Garmin Forerunner 970 while also syncing runs to Strava. Garmin predicted a 2:00:51 finish—faster than the runner’s previous personal best. Strava, by contrast, suggested 2:10:34, even slower than a prior official half marathon of 2:05. In other words, the two race time estimators disagreed by nearly ten minutes before the starting gun. Race day itself added complications. The course featured a helpful net downhill in the second half, encouraging fast times, but the weather turned significantly hotter than any recent training run. Officially, the runner finished in 2:04:49, splitting the difference almost exactly between Garmin’s optimistic projection and Strava’s conservative one. The outcome highlights how each platform’s underlying assumptions collide with real-world variables like heat, pacing execution, and race-day nerves.
Garmin: Optimistic Ceiling, Not Guaranteed Finish Time
In this head-to-head, Garmin clearly skewed optimistic. Its 2:00:51 half marathon call was essentially a best-case scenario based on strong VO2 max readings and recent training, not a guaranteed race outcome. Because the Race Predictor focuses on physiological potential, it often underestimates how long a race will actually take once imperfect pacing, fatigue, and environmental stressors creep in. Even Garmin’s more advanced course and weather-specific predictor—available when races are loaded into Garmin Connect on higher-end devices—cannot fully anticipate how a hot day or a shaky taper will affect performance. For runners, this means Garmin is best viewed as an upper limit: a glimpse of what might be possible on an ideal day. It can be highly motivating if you thrive on stretch goals and aggressive targets. However, relying on it as a precise race time estimator can set you up for disappointment if conditions or execution fall short of perfect.
Strava: Conservative Guardrail with Volatile Swings
Strava’s half marathon prediction of 2:10:34 undershot the runner’s actual 2:04:49 result, continuing a pattern of conservative estimates. Because its AI model weights long-term history and all-time efforts, it often anchors to the bulk of your training—which may be slower easy runs rather than race-specific sessions. That can lead to predictions that lag behind fitness gains, especially if you have recently introduced more intensity or are returning from a layoff. At the same time, many runners report that Strava’s predictions can feel jumpy, lurching up or down after just one unusually good or bad run. With no adjustment for course profile or altitude, it effectively imagines a flat, neutral track. The upside is that Strava tends to overestimate finish times, building in a margin of safety. If you prefer realistic baselines or want to avoid overreaching on race day, its cautious bias can act as a helpful guardrail.
Which Prediction Should You Trust for Your Training Strategy?
The Brooklyn Half experiment shows that Garmin vs Strava is less about which is universally “better” and more about which bias fits your mindset. Garmin’s running watch predictions are ideal if you want an aspirational target to build training around; its race time estimator reflects your potential ceiling when everything clicks. Strava’s performance predictions are better if you value conservative planning, using its slower estimates to shape pacing strategies and avoid going out too hard. In practice, many runners can benefit from using both. Treat Garmin as your stretch goal and Strava as your safety net, then aim to finish somewhere in between—just as the half marathon test did. Understanding each tool’s blind spots around heat, terrain, and training history helps you interpret their numbers, rather than obey them. Ultimately, neither app replaces thoughtful pacing, honest self-assessment, and race-day adaptability.