Instrument 01
Strava Plausibility Check: KOM Watts per Kg and a Cycling Verdict
Four numbers from any public activity, and a reading against the ceiling of recorded human performance.
A Strava plausibility check answers one narrow question: could a human being on a bicycle have produced this result? It does not read minds and it does not accuse anybody. It takes the figures a cycling activity already displays and returns the implied power, in watts per kg, that the effort demanded.
Run the Strava plausibility check
Reading
The tick is the effort. The heavy mark is the ceiling of published professional performance for this duration.
| Quantity | Value |
|---|---|
| Implied power | 0 W |
| Watts per kg | 0 W/kg |
| VAM | 0 m/h |
| Average speed | 0 km/h |
| Average gradient | 0 % |
| Duration | 0 |
| Ceiling for this duration | 0 W/kg |
How the Strava plausibility check reconstructs cycling power
Every watt a cycling effort produces goes to one of three places. Some lifts the bike and body against gravity, some overcomes the tyres rolling on the road, and the remainder pushes air out of the way. Add those three, divide by drivetrain efficiency, and the implied power is the figure the effort required.
On a climb the gravitational term swamps the other two, which is why a Strava plausibility check is at its sharpest on steep terrain. Four hundred metres of ascent has to be paid for in joules no matter how a rider is positioned, what tyres are fitted, or how kindly the wind happened to blow that afternoon.
Dividing by rider weight gives watts per kg, the figure that makes two cyclists of different sizes comparable. A 90 kg rider producing 400 watts and a 60 kg rider producing 267 watts are climbing at the same rate. Relative power decides a hill, and every honest cycling comparison has to use it.
The assumptions, stated plainly
An outside estimate has to assume the things it cannot see. This one uses a rolling resistance coefficient of 0.004, a drag area of 0.32 square metres, sea level air density of 1.225 kg per cubic metre, and drivetrain efficiency of 97.6 percent. Each is a reasonable road-bike default, not a measurement.
Those assumptions matter least where the check matters most. On a ten percent gradient the aerodynamic term is a small fraction of the total, so an imperfect drag guess moves the answer very little. On a flat effort the same guess dominates, and the resulting watts per kg should be treated far more loosely.
Reading a KOM verdict against the professional ceiling
Raw watts per kg means nothing without a duration attached. Sustainable power falls steeply as an effort lengthens, so the instrument compares each result against a ceiling that falls with it. Five minutes is judged against roughly 7.5 W/kg; twenty minutes against roughly 6.4; an hour against a little under 6.
When a KOM lands above that line, the verdict says so and stops. It does not name a cause, because from the outside the causes are indistinguishable. A motor, a mis-measured segment, a bad weight entry and a ride accidentally recorded in a car all produce numbers no cyclist could have generated.
This restraint is deliberate rather than timid. Segment leaderboards are full of results that look absurd and have entirely dull explanations, and full of results that look ordinary and are not. An instrument that reports the reading and declines to supply a motive is more useful than one that guesses at one.
Run the Strava plausibility check on a result you already believe, too. Seeing an ordinary club ride land mid-scale calibrates the instrument for you, and makes the reading on a disputed one considerably easier to interpret afterwards. A scale is only useful once you have watched it weigh something familiar.
Can you fake Strava data, and can a fake be spotted?
Yes to the first, and only sometimes to the second. An activity file is a text document describing where a device claimed to be at each moment. Nothing in the upload path verifies that the described ride happened, which is the structural fact every argument about segment leaderboards eventually runs into.
What a fake Strava file cannot easily do is stay coherent. Speed, gradient, heart rate and the shape of a climb all constrain one another, and an edit to one of them leaves the others contradicting it. Most attempts change a single channel and leave every other channel in the file arguing with it.
That is what this instrument tests. It does not hunt for a fake Strava upload, because tampering and a broken sensor produce exactly the same arithmetic seen from outside. It asks only whether the figures describe something a human could have done, and reports that answer without ever proposing a reason for it.
Strava plausibility check questions
Do I need the rider’s power file to run a Strava plausibility check?
No. The check works from the four numbers shown on any public activity: distance, elapsed time, elevation gain and rider weight. It reconstructs the power that those numbers demand. That is the point of the method, because the file belonging to a disputed KOM is exactly the thing an outsider never has.
How accurate is the implied watts per kg figure?
Treat it as a band, not a measurement. Rolling resistance, air density and riding position are assumed and not measured, and each shifts the result by a few percent. On a steep climb the estimate is fairly tight, because gravity dominates. On flat or descending efforts the aerodynamic guess carries far more weight.
What makes a cycling result implausible, not merely impressive?
Duration is what separates them. Almost anyone can exceed 7 watts per kg for fifteen seconds. Holding it for twenty minutes puts a rider above the published professional range. The check compares the effort against a ceiling that falls as the effort lengthens, so a short burst is judged differently from a long climb.
Does a flagged verdict mean somebody cheated?
It does not. A GPS error, a mis-set rider weight, a car journey recorded by accident or a segment whose distance is wrong will all produce impossible numbers honestly. The verdict says the figures cannot describe a human on a bicycle. Which of the several possible reasons applies is a separate question.
Can you fake Strava data, and would this check catch it?
An activity file can be edited, and nothing in the upload path verifies that the ride happened. This check does not detect tampering as such. It establishes whether the numbers describe a physically possible effort, which catches a crude fake reliably and says nothing at all about a careful one.
Can the check be used for running as well as cycling?
The power model assumes a bicycle, so the watts figure will not transfer. The VAM reading and the climbing rate remain useful for runners, because vertical metres per hour is a discipline-neutral measure. A dedicated running plausibility model is a different instrument altogether and is not what this particular page provides.
Related instruments: the watts per kg calculator for a single known power figure, the VAM calculator for climbing rate alone, and the GPX audit when you do hold the file. On the reasoning behind disputed results, see how KOMs are verified and why an activity gets flagged.