Normalized power for cycling explained

Average power is an easy metric, but normalized power tells a more complete story.

Member Exclusive

Become a member to unlock this story and receive other great perks.


Already a member?

Sign In

The act of riding, training, and racing a bicycle is a highly variable, or stochastic, exercise. There are many factors that affect every ride you take: wind, uphills, downhills, quick accelerations, long steady grinding, and so on. Because of this variability, average power is just not a sufficient indicator of the true metabolic demands of your ride. To account for this variability, we developed a special algorithm to calculate an adjusted (or normalized) power for each ride or segment of a ride (longer than 30 seconds) that you may want to analyze.

The algorithm incorporates two key pieces of information: (1) Physiological responses to rapid changes in exercise intensity are not instantaneous but follow a predictable time course, and (2) many critical physiological responses, such as glycogen utilization, lactate production, stress hormone levels, and the like, are curvilinearly, rather than linearly, related to exercise intensity.

To calculate Normalized Power:

  1. Start at the beginning of the data and calculate a 30-second rolling average for power.
  2. Raise the values obtained in Step 1 to the fourth power.
  3. Calculate the average of all the values obtained in Step 2.
  4. Find the fourth root of the number obtained in Step 3.

Many power-meter software analysis packages calculate Normalized Power for you. Basically, it’s the wattage you would have averaged if you had pedaled smoothly for the entire effort—the power that your body thought it was doing, though in reality the effort could have been a very sporadic on/off affair. It estimates the power that you could have maintained for the same physiological cost if your power output had been perfectly constant (such as on a stationary cycle ergometer), rather than variable.

Keeping track of Normalized Power is therefore a more accurate way of quantifying the actual intensity of training sessions and races. For example, it is common for average power to be lower during criteriums than during equally difficult road races, simply because of the time spent soft-pedaling or coasting through sharp turns during a criterium. The Normalized Power values for a criterium and a road race of about the same duration, however, will generally be very similar, reflecting their equivalent intensity. In fact, during a hard criterium or road race of about 1 hour in duration, Normalized Power will often be similar to what a rider can average when pedaling continuously for a flat 40 km time trial. The Normalized Power from mass-start races can therefore often be used to provide an initial estimate of a rider’s threshold power.

The figure below shows the difference between average power and Normalized Power in a road race. In this figure, the power line is constantly fluctuating, indicating that this section of the race contained times of high wattage and times of low wattage. This is typical of road races, where the range of power that cyclists produce is very wide and constantly changing. Since these changes in power output occur so quickly, the body does not have enough time between them to fully recover.

Graph showing comparison of normalized power versus average power in a mass-start road race

Thus, although the muscles get very short breaks, the body experiences the same amount of stress that it would if you did one hard, constant effort. Note that in the figure above the Normalized Power is 357 watts, whereas the average power is 319 watts. In this case the stress, or physiological cost, to the body was equivalent to what it would experience at 357 watts. The greater the difference, the more variable and less continuously aerobic the effort was. Charles Howe coined the term Variability Index to describe this. To find the Variability Index, simply take the Normalized Power number and divide it by the average power number. The more variable your ride (after the 30-second smoothing has been applied), the higher the Variability Index.

The reason all of this is important is that, used correctly, Normalized Power can help you to better define the demands of your event. The table below shows typical Variability Index values for some common types of cycling events. This is just a rough guideline for helping you to think more critically about variability in cycling. Knowing the demands of your event is one of the key factors to training specifically for that event. If you are a mountain biker and you are training only on the road, then most likely you will not be ready to handle the constant change in power, cadence, and speed that you will encounter in your next mountain-bike race.

Variability Indexes for Common Rider

Type of Ride Variability Index
Steady isopower workout 1.00 - 1.02
Flat road race 1.00 - 1.06
Flat time trial 1.00 - 1.04
Hill-climb time trial 1.00 - 1.06
Flat criterium 1.06 - 1.35
Hilly criterium 1.13 - 1.50
Hilly road race 1.20 - 1.35
Mountain-bike race 1.13 - 1.50

The table above shows Normalized Power and average power on a steady, relatively constant gradient climb. It is clear that this type of climb has a much smaller effect on the variability of a rider’s power output than the mass-start road race shown in the figure above. The wattage line shows how much smoother and more stable the power output was in this effort. The Normalized Power for this section of the ride was 304 watts; the average power, at 300 watts, was only 4 watts lower. Although these different efforts have roughly similar average power values, their Normalized Power and physiological cost are markedly different.

Adapted from Training and Racing with a Power Meter, 3rd edition, by Andrew Coggan PhD, Hunter Allen, and Stephen McGregor PhD with permission of VeloPress.

Training and Racing with a Power Meter, 3rd Ed.