How to use the normalized power and training load calculator
- Paste your power samples in watts, one number per sample. Write 150x60 for 150 watts repeated 60 times. Choose the seconds between samples.
- Enter your threshold power, the power you could hold for about an hour.
- Read the weighted average power, the plain average, the variability index, the intensity factor and the load score for the ride.
- Check the number of 30 second windows used, and the best 30 second average.
Formula
rolling mean over 30 seconds of the samples, starting when 30 seconds of data exist
weighted average power = (mean of the rolling means to the 4th power) to the power 1/4
variability index = weighted average / plain average
intensity factor = weighted average / threshold power
load score = duration (s) x weighted average x intensity factor / (threshold x 3600) x 100
The fourth power gives hard efforts more weight than easy ones, so a ride with surges scores higher than a steady ride with the same average.
Three minutes in three blocks
60 seconds at 150 W, 60 at 250 W and 60 at 100 W give a plain average of 166.7 W. There are 180 - 29 = 151 rolling windows. The mean of their fourth powers has a fourth root of 197.2 W, so the variability index is 197.2 / 166.7 = 1.183.
With a threshold of 200 W the intensity factor is 0.986. The load score is 180 x 197.2 x 0.986 / (200 x 3600) x 100 = 4.9. A steady 3,600 seconds at exactly 200 W gives a weighted average of 200 W and a load score of 100.
What the numbers describe
The plain average hides how uneven a ride was. Power from a bike is very variable, and the body does not respond to power in a straight line: short hard efforts cost more than their average suggests. Weighted average power estimates the steady power that would have a similar cost, which is why it is higher than the average for a variable ride. The variability index tells how uneven the ride was, and the load score turns duration and intensity into a single figure where one hour at threshold is 100.
To set training zones from threshold power, see the cycling power zones calculator. To fit a power model to your best efforts, the critical power model calculator is available, and the heart rate zone calculator covers effort by heart rate.
Limits and differences between programs
Different programs handle the first 30 seconds, pauses and dropped samples in different ways, so the same file can give slightly different numbers. This page needs at least 30 seconds of data, begins averaging once a full window exists and treats every sample as consecutive. Gaps should be filled with zeros or removed before you paste. The load score is a published training-load formula applied to numbers you enter. It describes the ride and is not a recommendation, a diagnosis or medical advice.
Assumptions and limits
- Samples are evenly spaced, one every 1, 2, 3, 5, 6 or 10 seconds, so a 30 second window is a whole number of samples.
- Powers are from 0 to 5,000 watts. Up to 100,000 samples can be pasted.
- Threshold power is a good estimate of your current one-hour power, and is an example value until you replace it.
- Rest and coasting count as zero watts if the file records zeros, and are skipped if it does not record them.
Frequently asked questions
Why is weighted average power higher than my average?
The fourth power in the formula counts hard efforts more than easy ones. A ride with steady power gives nearly the same value as its average, and a ride with surges gives a higher one.
What does a variability index of 1.00 mean?
It means weighted average power equals average power, which happens when power is perfectly steady. Most real rides land above 1.00, and rides with many surges or stops go higher.
What is a load score of 100?
It is the score for one hour exactly at threshold power. Shorter or easier rides score less, and harder or longer ones score more. It compares sessions on one scale and says nothing about recovery.
Can I paste a long ride?
Yes, up to 100,000 samples, which is about 27 hours at one per second. Use the 150x60 repeat notation for long steady parts to keep the box short.