Normalized Power: What It Is and Why It Beats Average Power
Normalized Power estimates what a variable ride really cost you, as if you had held a steady effort. Here is how it differs from average power, how it is calculated, and how it feeds intensity factor and TSS so your training load reflects reality.
Why average power lies to you
Imagine two one-hour rides. In the first, you hold a dead-steady 200 watts the whole way. In the second, you alternate between coasting at 0 and hammering at 400, averaging out to the same 200 watts. The average power readout says these rides were identical. Your legs know they were not. The second ride, full of hard surges and recoveries, cost you far more.
This is the problem Normalized Power exists to solve. Average power is the simple mean of every reading, coasting zeros included, and it quietly understates how demanding a variable ride actually was. Normalized Power estimates what really happened: the steady effort that would have produced the same physiological stress as your actual, spiky ride. On almost any real outdoor ride, it is higher than average power, and the size of that gap tells you how variable the ride was.
What Normalized Power represents
The core idea rests on a fact about physiology: your body does not respond to power in a straight line. Doubling your power does not double the stress, it more than doubles it, because lactate accumulation and fatigue climb steeply with intensity. Brief hard efforts therefore cost disproportionately more than the watts alone suggest.
Normalized Power bakes that nonlinearity into a single number. It answers the question "if I had ridden perfectly steadily, what power would have felt as hard as this ride did?" For a flat time trial, the answer is barely above your average power, because the ride was already steady. For a criterium full of accelerations out of corners, or a hilly group ride, the answer can be 20-40 watts higher, because all those surges carried a hidden cost that average power ignored.
How it is calculated
The standard calculation has four steps:
- Take a 30-second rolling average of your power data. This smoothing reflects how your body actually responds to changes in effort, which is not instant.
- Raise each of those 30-second values to the fourth power.
- Take the average of all those fourth-power values.
- Take the fourth root of that average.
The result is your Normalized Power. The two steps that matter most are the 30-second smoothing, which models your physiological response time, and the fourth-power weighting, which is what makes hard segments count for so much more than easy ones. It is a model rather than a physical law, but it lines up well with how fatigue accumulates in practice, which is why it has become the standard.
How to read the gap
The difference between Normalized Power and average power is itself useful information. Their ratio is sometimes called the variability index. A value near 1.0 means a steady ride, like a time trial or a disciplined indoor session. A high value means a spiky ride, like a race or a hilly group ride with repeated surges.
That has practical meaning. For a time trial or a long steady distance effort, you generally want low variability, because a smooth effort is the most efficient way to cover ground. Seeing a high variability index on a ride you intended to be steady is a sign you were surging unnecessarily and bleeding energy. On a race or a hard group ride, high variability is expected and simply tells you the day was punchy.
Where Normalized Power fits in your training
Normalized Power is not an end in itself. It is the input that makes the rest of the load model honest. Divide it by your FTP and you get your intensity factor, a clean measure of how hard a ride was relative to your threshold. Combine intensity factor with duration and you get Training Stress Score, the number that drives your fitness and fatigue trends.
That chain is only as good as its weakest link, which is your FTP. If your threshold is out of date or set too low, your intensity factor reads too high and your TSS inflates, and the error flows all the way through to your fitness chart. Keep FTP current with a periodic test or a confident estimate from recent best efforts, and the whole model holds together. The FTP calculator turns a test result into FTP, and the TSS calculator shows how Normalized Power and FTP combine into training load.
For most riders, the takeaway is simple. Stop judging variable rides by average power. Normalized Power is the more honest number for how hard you actually worked, it is the basis for comparing spiky rides to steady ones, and it is the foundation the rest of your data-driven training is built on.
