Should Event Categories Be Based on Power to Weight or FTP?

For flat and hilly routes, success in real life racing and group rides often depends more on raw power (FTP) than power to weight (w/kg). GCN has an excellent video on this.

If this is true on Zwift as well, should group events continue to be based on power to weight instead of FTP?

To answer this question, we need to determine the following:

1. Does power to weight create a significant advantage for heavier, more powerful riders?
2. Would FTP create a significant advantage for lighter but equally powerful riders?
3. Which method is better?
4. Are there alternatives?

Does Power to Weight Create a Significant Advantage for Heavier Riders?

We analyzed ZwiftPower race results by comparing the finishing order of 2,000+ racers in 17 Zwift races.  We grouped racers into power groups, recorded their power to weight and percentile finish in the race (lower is better) and charted the results (see below).  If power to weight is a fair method, you’d expect the graph lines to be flat indicating similar race results (Y axis) across all power groups (X axis). But, instead, at virtually every power to weight across every race category (A, B and C) the graph lines are descending, clearly indicating that the heavier, more powerful racers, have significantly better results.

Percentages are the average race result from the top for each power group (e.g. 40% means that, on average, that power group finished in the top 40% of all racers). Power groups represent the normalized power plus 20 watts (e.g. the 260 power group includes all racers with normalized power between 260-280 watts).

Speed/time tests were then conducted to determine just how much faster heavier riders are at the same power to weight. Three volunteers chose either a flat or hilly loop and performed three identical Zwift workouts at different weights and raw power but the same power to weight. Not surprisingly, over a one-hour event, the heavier riders were between 1:00-1:31 minutes faster for every 10 kg weight difference depending on various factors such as gradient, power and weight.

Based on this speed/time test and our weight study (see appendix), these results can significantly impact events.

1. In a 100-person event, a C rider of average weight is 10-20 kg lighter than approximately 16 people… a 1:00-3:00 minute time disadvantage in a one-hour event.
2. In a 100-person event, the lightest 16 C riders are 10-40 kg lighter than approximately half the field… a 1:00-6:00 minute time disadvantage in a one-hour event.

Would FTP Create a Significant Advantage for Lighter but Equally Powerful Riders?

Although events based on power to weight favor heavier riders, should events be based on FTP instead?  Not necessarily.  Since we expect lighter riders to be faster than heavier riders at the same power, we need to test this speed advantage and compare the results to the power to weight test.

For the FTP test, three volunteers once again chose either a flat or hilly loop and performed three identical workouts at different weights but the same raw power.  As expected, over a one-hour event, lighter riders were between 1:35-2:22 minutes faster for every 10 kg weight difference.

Which Method is Better?

In the power to weight test, heavier riders were between 1:00-1:31 minutes faster for every 10 kg weight difference. In our FTP-based test, lighter riders were between 1:35-2:22 minutes faster for every 10 kg weight difference.

Based on these results, it becomes clear that events based on power to weight give an advantage to heavier riders, but events based on FTP would give a greater advantage to lighter riders.

Are There Alternatives?

So, it turns out that power to weight, for all its faults, is slightly better than FTP.  Is there a more effective method for closing the wide time gaps we see during events?  Below are the pros and cons of two alternatives:

1: Hybrid Power to Weight

Same as currently but with minimum and maximum FTP limits. Riders above the maximum move up and riders below the minimum move down.

Pros:  Shorter event time gaps; easy to implement.

Cons:  At the same power to weight, the remaining heavier riders still have an advantage.

2: Results-Based

You race, you get results and you are placed in a category based on those results.

Pros:  A fair, merit-based system which is used in real life racing.

Cons:  Someone has to create and administer a system that has the backing of Zwift and the entire Zwift community.

Regardless of some imperfections, Zwift events are still the backbone of the platform which has made indoor cycling not just tolerable but social, fun and addicting. Hopefully as the platform matures more effective event classification can be implemented, resulting in a better experience for all involved.

Appendix

Fun facts about weight from our study of 2,454 Zwift racers (A=629, B=894, C=728, D=203… too few to be included in our race results analysis):

1. The average Zwift racer weighs 76.8 kg (169.3 lb) with a standard deviation of 9.2 kg (20.3 lb)
   At one standard deviation, 68% of the racers weigh between 67.6-86.0 kg (149.0–189.6 lb)
   At two standard deviations, 95% of the racers weigh between 58.4-95.2 kg (128.8–209.0 lb)
2. “A” racers weigh the least and have the lowest weight variation. Each successive group weighs more and has a higher weight variation.
   The average A racer weighs 72.3 kg (159.4 lb) with a standard deviation of 7.4 kg (16.3 lb).
   The average B racer weighs 75.7 kg (166.9 lb) with a standard deviation of 8.6 kg (19.0 lb).
   The average C racer weighs 79.7 kg (175.7 lb) with a standard deviation of 10.4 kg (22.9 lb).
   The average D racer weighs 85.7 kg (188.9 lb) with a standard deviation of 12.9 kg (28.4 lb)
3. In Zwift races, the average weight difference between the heaviest rider and lightest rider is 44.1 kg (97.2 lb).
   The average minimum weight in a Zwift race is 55.7 kg (122.8 lb)
   The average maximum weight in a Zwift race is 99.8 kg (220.0 lb)

Fun facts from our study about normalized (weighted average) power:

1. The average Zwift racer puts out 269 watts with a standard deviation of 34 watts.
   At one standard deviation, 68% of the racers put out between 235-303 watts.
   At two standard deviations, 95% of the racers put out between 201-337 watts
2. Each category averages about 40 watts more or less than its closest category. All of the categories have about the same power variation.
   The average A racer puts out 318 watts with a standard deviation of 36 watts.
   The average B racer puts out 277 watts with a standard deviation of 35 watts.
   The average C racer puts out 237 watts with a standard deviation of 35 watts.
   The average D racer puts out 197 watts with a standard deviation of 34 watts.
3. The average minimum and maximum power for all races is 190 and 343 watts, respectively, for a difference of 154 watts.
   The average min and max power in an A race is 244 and 398 watts, respectively, for a difference of 153 watts.
   The average min and max power in a B race is 189 and 352 watts, respectively, for a difference of 163 watts.
   The average min and max power in a C race is 160 and 310 watts, respectively, for a difference of 150 watts.
   The average min and max power in a D race is 127 and 257 watts, respectively, for a difference of 130 watts.

Fun facts from our study about race times. Keep in mind, these results are AFTER Zwiftpower disqualified or moved riders up from the real-time Zwift results.  In addition, we eliminated the bottom 5% of racers in each race Category to prevent severely lagging racers from skewing the results.

1. The average Zwift racer finished about 3:20 minutes behind the winner with a standard deviation of 2:03 minutes.
2. In each successive category, the average racer finished further behind the winner.
   The average A racer finished 1:30 seconds behind the winner.
   The average B racer finished 3:01 seconds behind the winner.
   The average C racer finished 4:50 seconds behind the winner.
   The average D racer finished 6:19 seconds behind the winner.
3. On average, the last place racer finished 8:17 seconds behind the winner.
   The average last place A racer finished 4:54 behind the winner.
   The average last place B racer finished 7:16 behind the winner.
   The average last place C racer finished 10:07 behind the winner.
   The average last place D racer finished 16:44 behind the winner.