1ST 2ND
Cat A 3.6w/kg 3.7
Cat B 3.7 3.6
Cat C 3.6 3.4
Cat D 3.4 3.3
Cat E 2.8 2.9
Posted without comment.
Without comment, context, or utility.
1 Like
The stats speak for themself, what are you finding hard to comprehend?
This doesn’t really speak to how the race played out. It doesn’t show watts, just w/kg. There isn’t any recognition of pack size, effort levels, or otherwise. Was it 2 people or the whole pack. Were there only 2 in A, were people further down with higher.
What was the point you are attempting to make?
Here’s a 300 person time slot. ZwiftPower - Login
A better way to get context out of this;
- look at the KOMs to see how hard people punched.
- Look at the normalized power for categories to see how stop/start the racing was or if it was a GC focused group
- look at how the skilled riders who won with way lower w/kg than (a) they were capable of and (b) those behind them.
In this case you can see the As were pretty slow over the KOM 1, but very punchy. High normalised vs total watts.
The Bs were faster on Titans and I’d assume the front group split off and hit and gap before doing more like A. The second group of Bs was a steadier effort.
C and D are more of a flat effort throughout.
Have a look for yourself, all decent size fields, Cat E won by a Cat C, Cat D won by a Cat B, Cat C won by a Cat B and Cat B won by a Cat A and Cat A won by a Cat A.
The point am making is the first 3 or 4 in Cats ABC are closer together in racing terms than most of the riders in their pens they are racing against yet they are in 3 seperate pens, ZRS was meant to be fairer, its a total failure!
1 Like
If these were rebranded CE the Cats would be A, B+, B-,C and D+
The A’s were not really trying. 20seconds off their best for Titans (same riders) and 20 seconds off their best for forward KOM. Werme was 100w and 1.2w/kg off his PR.
Similar story in B with the top 2 particularly, but most of the top 10.
Winner of C should be higher cat. 2nd had no race history. 3rd was super low wattage 227 for the w/kg. You can see that 5% variability is common in C vs the 15% in A, so the effort is flatter and likely close to best 20min.
And then D. Two lighter side riders at the top doing a flat effort for them with no variability. Probably graduate to C and get smoked based on results if they pick punchy/rolling races.
So yea, looking across categories. A and B phoned it in. A couple of people in C and D are ranked a bit low on ZRS but it is due to lower raw watts and 30sec. Raw watt wise the winner of D is 15th percentile across the population so on a 5 way split of categories it kinda makes sense they would be in D, which is a shame because they got upgraded to C.
I’d say ZRS is working ok but has some potential tweaks to seed scores for riders that haven’t sprinted.
“I’d say ZRS is working ok but has some potential tweaks to seed scores for riders that haven’t sprinted.”
Please tell me what metric are you using to come to this conclusion ?
We have created an extra Cat so the difference between top half (for example average w/kg of top 5 and bottom 5 riders) and bottom of a race in w/kg should have come down but looking at most races the difference has went up by quite a bit so I can clearly say the races are less fair as there is a bigger spread of abilities in each race, if any of this is wrong please point it out.
There are persistent category challenges and ZRS-specific categorization challenges. Zwift has bravely or foolishly taken on most of them at the same time by switching the system, as the number of riders coming back to the platform after 90 days is at its highest.
If someone comes back onto Zwift, does a couple of warm-up rides, and then jumps into a race, they will be miscategorized under both systems. They could choose to enter based on their perceived ability, but as we’ve seen, a lot of folks just jump on the lowest they can. Under CE, people with high 20-minute scores would be catapulted up the cats very quickly. Under ZRS, this only happens if they hit both 30s and 10-minute scores.
Similarly there are a lot of accounts on really junk power inputs. Including now spin bikes through the QZ app. They tend to have “zpower” like profiles of a high consistent wattage but lack the resistance to go harder for sprints. They were a nuisance in CE and are a nuisance in ZRS… to a different category.
The primary flaw of CE was that heavier or skillful (sandbaggin?) riders with low w/kg would dominate lower cats with an easy ride then unleash a huge sprint that doesn’t cat them up. The flaw under ZRS is the opposite. There is a lot of churn initially and people seem pretty emotional about the churn / loss of status.
A great example of how this is starting to be more realistic cat progression is the Rhino chase race. This was the first week in a long time that the A’s won, and they won really easily, catching the B group halfway. This suggests the cats are pretty well set up if all cats race the same way (in this case all out).
You attempted to analyze the w/kg under the assumption that all races were approached the same way, which they were not. Using total watts on a short course with 2 punchy features is more likely to be off. As I showed above the race was approached more tactically in A and B, and more like a TT in C and D.
Is it going to be fairer in the long run? I think yes because there is a recognition that the game is not just about a raw 20min w/kg. People will gain ZRS if they are good at the gameplay + have power. People will lose it if they lack either. This gives progression and stops the same folks winning and staying in Categories race after race all winter.
With cheats and categorization, we all have a choice: focus on what we can control and get stronger or worry about what is out of our control. We can also give zwift nudges on where issues are, but to do that effectively requires nuance and detail.
The necessary tweaks that I see are regularly looking at population by cat, initial seeding (30 seconds), and speed of progression for people with very high 10min and very low 30s. These can be quick alterations to the algo that have a big impact; slow change might be best.
4 Likes
Again unless you have a metric that you are using to measure fairness then I really dont see how you can have a valid opinion.
I am very sorry that this race you didn’t compete in was so unfair. I hope it doesn’t detract from your season on Zwift.
3 Likes
Thanks for your unscientific input, btw if ZRS seed scores dont correlate to ability which they don’t then there is no way ZRS can work!
Work meaning, to be fairer than the system it replaces.
Average power for the whole race is a terrible measure. On that course 2 minute power is where the race is decided (two climbs of around 2 minutes).
Here’s the max and median (approx median, I’m not going to spend time counting where the exact middle placing is) 2 minute W/kg for the race posted:
A: 6.6, 5.7
B: 5.9, 5.2
C: 5.6, 4.4
D: 4.8, 3.4
E: 3.5, 2.6
The medians have quite an even distribution but you could certainly argue that the top in C and D (and maybe E) are quite a way above the median in their categories. In fact, the top in B through E are all above the median in the category above.
Isn’t that an undesirable outcome? If the top in one category is better than the median in the category above then maybe they should be in that category above?
5 Likes
Good way of looking at it based on the reality of the race.
Messier to get the data for what they could have done if all out.
The top 3 As have 5min PRs above the median 2min. (3rd place set their 5min PR in the race). They were chilled out up the climbs. The A race was decided by two all out sprints in the last 1.5km.
Seems like the top 4 Ds set 90 day PRs from at least 2min to 5min.
1 Like
“If the top in one category is better than the median in the category above then maybe they should be in that category above?”
Exactly this, there is too much of an overlap of abilities in each Cat, There is 5 pens now so Cat E should in theory have riders of a lesser ability than old Cat D but this is not the case.
1 Like
The top people in 2 cats were upgraded. So yes… maybe they should be.
A 23min flat race in CE last year before ToW started. 20min values. zid=3812912
A: 4.0, 3.5
B: 3.9, 3.35
C: 3.9, 3
D: 2.5, 2.3
A punchy race with a 2min pinch. 2min values zid=3147060
A: 7.1, 5.8
B: 5.6, 4.8
C: 5.1, 3.8
D: 3.3, 2.9
It doesn’t feel notably different. D’s might have it better under CE using this tiny sample. The assumption across this thread’s analysis is that we don’t know how many people were not trying as hard as they could. It is clear that most in A were not across all of the races. B probably could have done more.
In terms of “best effort” and overlap. Walk through the top-ranked individuals in CE categories, and you will see plenty of your “higher than median in above” folks across multiple power metrics.
So here is the 2nd ranked C rider under CE.
And the 4th ranked C rider under CE.
top ranked B
2nd ranked B
ZRS at least has a way for these folks to get upgraded. They were comfortable in most CE races and didn’t need to hit their best numbers to win. I was classified as a B for most of the year because I had no reason to hit the 5-minute w/kg trigger. The fact that I could have meant I was more comfortable with the power numbers that I did put out.