What is Normalized Power vs. Average Power?

For those who train with a power meter on their bike – and if you don’t train with power, you should – doing a post-workout data analysis is almost as much fun as the ride itself. But if you don’t know how to interpret the power data you’re collecting, you’re not going to reap the full benefits of this training tool. 

Power meters come in different forms (for instance, in the pedals or crank arm) and measure a cyclist’s power output in watts as the result of the torque (force on the pedals) multiplied by your cadence (RPM, or revolutions per minute). Unlike other training metrics, like heart rate and miles per hour, which can be affected by outside factors like weather, terrain, or fatigue, power is absolute. It’s the real-time data on the actual work being done. 

Power is used to determine your Functional Threshold Power (FTP), which can be used to set training zones and determine targets for workouts and races. But FTP and real-time power aren’t the only terms you should know. There are a number of terms and data points to consider when using power: normalized power, average power, variability index, and intensity factor.

Average power vs. normalized power

The two power terms that come up most frequently when looking at your data post-ride are average power and normalized power. 

Average power is simply an average of the watts produced during a workout – including zeros. The only downside with this metric is that all power values have the same weight, so periods of lower output, like coasting, drafting, or stopping, can dramatically reduce the average. That’s where normalized power comes in. Normalized power is an estimate of the power you could have maintained for the same workload at a steady effort. It de-emphasizes periods of lower power output and places greater weight on harder efforts. 

“Consider a rolling course,” said Jeff Bowman of Rev Tri Coaching. “Most people will push power going uphill, crest the hill, and then coast a little going downhill. That’s why you will see a difference at the end of the ride between your normalized power and average power. Normalized power is more accurate to what your body is actually feeling.”

Contrary to popular belief, normalized power doesn’t simply take out the zeros to improve the average. Instead, the training software, for instance, TrainingPeaks, uses a fairly complicated, behind-the-scenes, four-step algorithm to calculate normalized power.

As an example, consider two different rides lasting one hour. The first ride is a steady state workout on a trainer, resulting in an average of 150 watts. Now, think about a one-hour ride on a hilly course with climbs that cause power spikes of 250 watts for minutes at a time, followed by periods under 50 watts when riding downhill. The average power for that ride might still equal 150 watts, but the second ride with periods of high intensity will certainly feel much harder than the steady-state workout. Normalized power will reflect these fluctuations in power and can, in some cases, provide a more accurate representation of your output.

RELATED: How to Get The Most Out of Reading Your Power Files

When to use normalized power vs. average power

There are situations when average power and normalized power will be fairly similar, like during a time trial on a flat course. But for a ride that has fluctuations in intensity, like on a hilly course, normalized power can be a more reliable metric.

Normalized power is very useful for triathletes during a race, especially on a hilly course. While many people have their bike computers set to display either 3-second or 10-second power, Bowman suggests adding another data field: normalized power. 

“For example, if I give an athlete a wattage range for a race, I will tell her to set her bike computer to display normalized power as well,” he said. “I always want to see the cost of the ride in terms of the big picture and not just in 3 to 10-second increments. You can spike your power up hills, within reason, but you want to keep your normalized power within a certain range for the entire race.”

By looking at multiple data fields, you get real-time data and a better idea of how you’re doing overall. One suggestion from Bowman is to check your normalized power about 30 minutes into a race to see if you’re staying within your prescribed range. If you’re riding a little hot and the watts are too high, you may need to pay more attention for the rest of the ride to recoup some energy before the run. 

There is one scenario when normalized power isn’t as helpful – on any effort shorter than 20 minutes.

How normalized power affects other performance analysis

Normalized power also plays a part in determining other performance analysis features. “Normalized power is input into the intensity factor and Training Stress Score [unique to the TrainingPeaks platform], which help indicate how hard a workout or race was,” Bowman said. 

TrainingPeaks uses normalized power to calculate your Training Stress Score (TSS), which is an estimate of the overall training load and physiological stress created by a workout. (For reference, riding all-out for an hour is approximately a 100 TSS.) 

Other metrics to know include: your intensity factor and variability index. Your intensity factor (IF) is the ratio of normalized power to your functional threshold power (FTP) – i.e., how hard this ride was for you, individually. For example, an easy ride might have an IF of .75, and an interval workout might have an IF of .95. 

A related metric is variability index (VI): The ratio of normalized power to average power, which provides an indicator of output variability. A VI of 1 means NP = AP, or a steady effort. In a triathlon, which is generally a steady state effort, the ideal VI is typically 1.05 or less.

By using normalized power in conjunction with other power data and training metrics, like heart rate, speed, and cadence, you’ll be able to make the most informed decisions when training and racing.