Punch your last half-marathon time into a pace calculator. It spits out one number for 26.2 miles. That number is often wrong by ten minutes or more. It happens to about half the runners who use it.
AthleteOS skips that guess. It rebuilds your marathon forecast every week. The forecast comes from what you actually trained, not a race you ran months ago.
Why the Riegel Formula Predicts Your Marathon Finish Time Wrong
Most online calculators run on a formula. A runner named Pete Riegel published it in 1977. It looks like this:
Marathon time = Half-marathon time x (26.2 / 13.1)^1.06
The 1.06 is a fatigue exponent. Riegel built it from race data. It spans everything from a sprint to an ultramarathon. It’s one average, applied to everyone. It doesn’t know if your fitness is rising or falling right now.
A 2016 study tested it on 156 real runners. Each runner had a marathon and two shorter races on record. Riegel’s formula missed by ten minutes or more. That happened for roughly half of them.
In short: the coin-flip odds on a Riegel prediction are worse than you’d guess.
Add a runner’s own race history and the error drops by nearly half. That’s the whole point of this chart. More real data beats one frozen number.
What Your VDOT Table Doesn’t Know About You
Daniels’ VDOT tables work differently. VDOT is a single fitness number. Jack Daniels calculates it from a recent race time. Then he maps it to equivalent times at every other distance.
Here’s the catch. Daniels’ own guidance says the marathon prediction only holds up under one condition. You need 12 to 16 weeks of matched long-run volume first. Skip that base and the VDOT number stops being a forecast. It becomes a ceiling you probably won’t touch.
No calculator asks whether you ran those long runs. It just assumes you did.
| Method | What It Uses | Updates | What the Research Shows |
|---|---|---|---|
| Riegel formula | One past race time | Never | 10+ min too fast for ~50% of runners |
| Daniels’ VDOT tables | One past race time | Never | Assumes 12-16 weeks of base already built |
| Regression from race history | Half-marathon time + weekly volume | Only when you race again | MSE 208-227 vs 380 for Riegel |
| Durability-adjusted models | Heart-rate drift during the race itself | After the race is over | Cuts error from 6.45% to 5.16% |
| AthleteOS marathon forecast | Fitness score, fatigue score, sessions | Every week | Recomputed from live training, not a fixed input |
The Predictor Already Sitting in Your Training Log
A 2017 study looked at 84 amateur marathoners at the Madrid Marathon. It found something calculators ignore completely. Your half-marathon time correlates strongly with your marathon time (r=0.812). Weekly training volume matters too (r=-0.479). More volume tracks with a faster finish.
Translation: what you’ve actually been running is a better predictor than any table lookup. A calculator can’t see your training log. A platform that ingests it can.
Durability: The Marathon-Specific Fade No Calculator Can See
Here’s the piece every static formula misses. The marathon has a physiology problem shorter races don’t share. Your heart rate starts climbing relative to your pace partway through. This happens even when effort feels the same. Researchers call this decoupling. It’s the plain-English cousin of what AthleteOS shows you as your drift ratio.
A 2022 study tracked 82,303 recreational marathoners. Decoupling kicked in around 25 kilometers into the race, on average. Runners who decoupled less finished dramatically faster.
Two runners can post the same fresh 10K time. One holds together late in the race. One doesn’t. A fresh-state formula can’t tell them apart. It never checks what happens after mile 15. That’s why durability research on why VO2max and lactate threshold both fall apart after 90 minutes matters so much for marathon pacing. A marathon behaves differently than a 10K.
Fresh fitness is the engine spec sheet. Durability is whether the engine still runs the same after two hours on the highway in July.
Your Taper Is Worth Real Minutes, and a Static Number Never Sees It
A taper is the planned drop in training volume before race day. A well-run one is worth real time. A study tracked 158,117 Strava marathon finishers. A disciplined 3-week taper produced a median gain of 2.6%. That’s about 5 minutes and 32 seconds for a typical finisher. The gain differs by sex too. Women showed a 3.12% median improvement with a strict 2-week taper. Men showed 2.14%.
None of that shows up in a calculator you ran back in January. Your taper hadn’t even started yet. AthleteOS’s race readiness check tracks your form as the taper actually unfolds. For more on how taper length interacts with your fitness curve, see the taper-length research built for 70.3 athletes. It applies the same volume-cut logic at a shorter race distance.
A number you calculated once can’t see a taper that hasn’t happened.
Fitness Score Minus Fatigue Score: The Model That Updates With You
This part of training science actually behaves like a forecast. Not a photograph. Think of your fitness score as a bank balance. It builds up slowly over months of training. Your fatigue score works more like this week’s spending. It rises fast and fades fast. What’s left over is your form. Fitness minus fatigue. It’s the plain-English read on whether you’d race well today.
Research on this model shows something important. It fits well across a group of athletes. But it’s genuinely unreliable as a one-time snapshot for one athlete. Studies put the time-to-peak-performance window anywhere from 25 to 61 days. That’s a wide range. You’d only trust a number like that if it gets recalculated constantly, not fit once and left alone.
Take a runner we’ll call Daniel. He’s 34, training for a fall marathon. His base came from structured Zone 2 volume. In week 8 of his build, his fitness score was still climbing. His forecast sat at 3:41. By week 14, his long runs got stronger. His taper got cleaner too. The same forecast had moved to 3:29. Twelve minutes faster. He didn’t run a single new race to earn it. A static calculator anchored to his week-8 half-marathon time would have missed every bit of that change.
Your forecast should move exactly as fast as your fitness does.
How AthleteOS Predicts Your Marathon Finish Time Every Week
AthleteOS doesn’t run the Riegel formula once and call it done. It recalculates your marathon forecast on a weekly cycle. Three live inputs drive it: your current fitness score, your current fatigue score, and your actual completed sessions. That includes whether your long runs and weekly volume are tracking toward marathon-specific readiness.
That structure catches exactly what a fixed formula structurally can’t. Fitness that changed since your last race. A taper that hasn’t happened yet. Durability risk that only shows up when your real long-run log gets checked against your actual race distance, the same idea behind threshold pace durability research.
A photograph tells you what was true the moment the shutter clicked. Your training doesn’t stop moving after that. Your forecast shouldn’t either. Want to see your own number update as your training does? Start tracking your marathon forecast with AthleteOS.