Upload your race course to estimate your bike split and build a course-specific pacing plan using your power, weight, bike setup, elevation and race-day conditions.
Pin exact course sections to a target power. Variable pacing adjusts the remaining course towards the overall NP goal; max-power and enabled duration caps still apply.
Save, reopen, duplicate and compare race plans locally. Uploaded GPX text is stored in your browser with the plan.
The calculator will translate the physics result into a practical race-day pacing recommendation.
The app is assessing the quality of the course, aerodynamic inputs, rolling-resistance assumptions, weather and technical-course modelling.
This is a practical planning envelope, not a statistical probability interval. Higher scores mean more real, race-specific inputs are supporting the model, so the planned split is more likely to finish close to the actual ride. The time range shows how much the estimate can move under reasonable changes to the least-certain inputs.
Save multiple named scenarios from this course and identify the fastest option that still looks raceable.
| Scenario | Type | NP | Split | Δ current | IF | TSS | Risk | Safe? | Change |
|---|
Explore the route by gradient, target power or predicted speed. Hover the map or elevation profile to inspect the same course point.
A practical execution plan generated from this exact course, power target and weather simulation.
| Scenario | Watts |
|---|
| Scenario | km/h |
|---|
Download course-specific target-power files for indoor platforms, head units or coach review.
Creates a planned-workout .zwo from distance-based course chunks. Exact manual kilometre boundaries stay separate; optimiser sections use sensible 1, 2 or 5 km chunks based on course length.
.zwo.TrainingPeaks currently imports planned workouts only from .zwo, so the library stores these distance-built chunks as predicted-duration blocks. On course, follow each interval's kilometre boundary and watt cue if your actual speed differs. Direct send is unavailable without approved TrainingPeaks developer API credentials. Official import instructions.
| Km | Grade % | Wind m/s | Power W | Speed km/h | Split | Elapsed |
|---|
What it does: full physics forward model (aero drag, rolling resistance, gravity, drivetrain loss) solved per course segment from your GPX, with per-segment wind from the forecast bearing and hour-shifted conditions as your predicted ride progresses. Power is stepped — steady on the flats, lifted by your climb boost on grades over 2% — so NP, VI and TSS are computed from the actual power distribution (duration-weighted, a close approximation of the rolling-30s method). The calculator now reports model version, calculation time and input confidence; broad public validation should be treated as in progress until a larger predicted-versus-actual dataset is published.
What it doesn't do: variable mode is a simplified optimisation rule — power scales with how slow each segment is, capped at your Max %FTP and rescaled to hit the NP goal. It captures the big idea behind a BBS-style plan (spend watts where they buy the most time) without claiming BBS's segment-by-segment optimiser or its decade of calibration. The advanced corner/braking model is still a planning estimate based on GPX bearing, speed and handling assumptions, not a replacement for course reconnaissance or safe riding. CdA and Crr are presets unless you've tested; a wrong CdA moves the answer more than anything else on this page. Cheat-sheet numbers are duration-weighted averages per scenario on this course in these conditions — treat them as anchors, not laws.
The rule: treat the range, not the number. This is a sanity-check and pacing anchor — race-day execution still comes from your MP race plan.
Use a clean ride with reliable power, speed/GPS, matching equipment and recorded conditions. The estimate is shown with course match and residual error; nothing is applied until you choose Apply. Traffic, braking, drafting, position changes or uncertain wind can bias CdA, so record them in the ride notes.
This report builds a predicted-vs-actual evidence base before any strong public accuracy claim is made. Add real race or ride outcomes, separate measured-CdA and preset-CdA cases, and explain large misses.