How predictions work

The marathon projection formula

Time projections use a formula developed by Vickers and Vertosick, published in 2016 in BMC Sports Science, Medicine and Rehabilitation. They analyzed thousands of race results and found that marathon performance is best predicted by combining a recent race time with weekly training mileage — not pace alone.

The formula converts your input race into an equivalent marathon fitness estimate, then adjusts it based on how much you're currently running. Higher mileage generally means better marathon-specific endurance: going from 30 to 50 miles per week typically shaves a few minutes off the projection, all else equal.

Training paces

Once the formula produces a marathon time, the app converts it to a VDOT score using Jack Daniels' method from Daniels' Running Formula. VDOT is a shorthand for your current aerobic fitness level, and Daniels' tables map it to training zones — easy, marathon pace, tempo, intervals, and repeats.

Course conversion

The course converter estimates how a finish time from one marathon translates to another. Each course has a difficulty multiplier calibrated relative to Berlin — the flattest, fastest reference point, and the consistent home of world records. Your time is scaled by the ratio of the two multipliers.

Weather adds a further adjustment using a wet-bulb globe temperature estimate (Stull 2011), with a performance penalty per degree above 10 °C WBGT (Ely et al. 2007, Medicine & Science in Sports & Exercise). Cold conditions are treated as optimal — the cold performance literature is murkier, and under-dressing adds its own variable.

What it doesn't know

The formula can't see your injury history, how long you've been at your current mileage, your sleep, or whether your input race was a genuine effort. A sandbagged 5K or a blown-up half marathon will skew the output. Weekly mileage should reflect a real recent average, not a peak week.

The model also doesn't account for altitude, wind, race-day fueling, course-specific pacing demands, or how efficiently you personally respond to marathon training. Two runners with identical inputs can have meaningfully different race days.

Use the output as a calibration tool and a starting point for race planning — not a guarantee.