I tried a test event today. I normally finish bottom half in C races. But the category enforcement put me in B which made me nervous. Turns out I was right be nervous as I got destroyed and finished dead last (not including a few people who didn’t finish). Looking at the class C results for the event I would have been close to the back there as well just based on times.
My average power was not up to my best. Maybe because I over did it in the beginning. After 3 minutes I was in a small group of three at the back already 30 seconds behind the next riders, not sure how far behind we were from the leaders. It was a small field though, I think something like 26 started.
As others have stated I think it might be a good idea if Zwift took into consideration you past results when selecting a category.
This is why ranking systems generally decay rather than deduct.
If you do comparatively better, your Ranking improves.
If you do comparatively exactly as before, your Ranking stays the same.
If you do comparatively worse than before, your Ranking stays the same, but decays over time.
You can see this even in the ZP algorithm. You don’t get negative gain, you just gain nothing.
It doesn’t solve all things, and presents problems of its own (relegation happens a bit later than might be optimal), but on the whole you’re not constantly bouncing up and down. Assuming brackets are stableish.
Fwiw, my reading of Bailey between the lines is that perfecting the power curve algorithm is the goal here. So I guess carry on talking about how the incorrectly categorizing system categorizes incorrectly, as it will forever do, and I’ma go look for a different platform that takes racing seriously.
I did my first races under this system. I am currently category D and it was lovely to actually race against the people in my category on the course as opposed to looking up my real competition on Zwift Power and racing them and ignoring most of the category D entrants.
I pushed much harder than normal so I suspect that soon I will move up to category C, where I will be at the back. At least the front runners will not be impossibly far away and I will be in the same race, albeit at the back!
I think that most people in Zwift simply don’t understand what their actual category should be. Some probably know that they are at the threshold for a higher category but are choosing to race easier. A very small number of people know that that are in races well below their ability.
This is a good first effort and I am sure that Zwift will learn a lot from it.
I would like to ask a question: I sometimes do weight-adjusted handicap races with members only (never at official events, of course), such as meetups, but will this affect the category judging?
I think it is one of the joys of being virtual that you can compete with anyone by increasing your weight even if you have a difference in leg strength.
I am aware that this may be seen as a gray area since the weight is set so far from the real world, but I feel that it would be a shame if this way of having fun disappeared.
In my opinion, it would be best to only look at watts, not w/kg, in the estimation phase of CP, VO2max, etc. and divide it by the weight at the time of race participation. Since body weight can fluctuate daily.
I’d like to see Zwift explore the idea of handicapping. All this talk about top of category and bottom of category and what category should you be in could be solved with a well designed handicapping system.
I would say you are mixing apples and oranges and creating a confusing situation.
You are not creating a group race, you are creating a group ride.
The moment the word race, in any incarnation is used, the entered info should be accurate.
To do otherwise creates chaos like we just saw with the Ralpha series.
Races and rides are 2 different activities.
Long awaited very cool! The categorisation parameters seem sensible and it’s definitely a huge improvement over 20min power only…
Now we only need race organisers to come to their senses and actually separate the different categories, and we’ll be making some steps towards a meaningful system of distinct race categorisation, in which riders at the lower end of the category (at least in categories D, C and B) can also enjoy a race.
I know a group that does this too and it leads to good tight races between people of vastly differing abilities. When you say it is common in the Japanese community, is that just in Zwift (or other virtual apps) or in real life too? If in real life how is the handicap applied?
Hmm, which of these placed 12th in A’s in a recent category test event. Which two placed 1st and 2nd in B’s, in the same race.
There seems to be a lot of attention paid to 1, 5 and 20 minute power.
15sec, possibly not so much.
Speaking for all time-trialists, yes, we can hang with faster groups longer. But unless we do happen to get off the front we rarely if ever win or podium if we end up in a bunch sprint even in lower categories. While we have all the sympathy in the world for people getting dropped mid-race, it seems there is little for those of us doomed to start every sprint towards the front and finish near the back.
Add 15sec into the model, or if there already increase the bias for pushing good sprinters up.
How exactly do you think 15 sec power should be added to the model?
We don’t know exactly how the 1, 5 and 20 minute power bands are used in the current model.
You could (for example) just use high 15sec power as another way to push people up a category. Set maximum’s for D, C, and B.
Alternately, looking at all four bands, you could move people down when they have a distinct weakness. E.g. we know that 1 min power is an absolute guide to how well you can hang on to Zwift race starts. And we also know that 15sec power is an absolute guide to how well people will do in final sprints.
Yes. Yes we do. Arrrggghhh. So much talking, so little reading.
You really can’t. Sprint ability can be very similar for cat Ds as cat As. They are very very very different systems. Riders have to be able to get to the end to have a chance to sprint, including sprint values is incredible difficult.
Track sprints can sometimes have sub 200w FTPs and put out 2500w.
I would assume that since we are using the CP model from power data it includes all data so any peak will influence the CP model curve. So it is a lot more than just a few points.
Track sprints can sometimes have sub 200w FTPs and put out 2500w.
I think you are making my point. Do you think that someone with a sub 2.4W/kg 20 minute power that can do a 2500W sprint should be allowed in the D’s? Or even C’s?
very cool! The categorisation parameters seem sensible and it’s definitely a huge improvement over 20min power only…
