Great, but you can show that by opening up the information flow more. Some of us have been discussing and analyzing this problem since about 2017.
I did both of the first two auto-cat tests. I didn’t do the first two official races as they don’t fit my profile (older time trialist that can’t keep up on hills longer than 1-2 minutes.)
I have signed up for this week’s Crit City race. The auto-cat process has downgraded me from C3 to C4. For a course and distance, I usually do well in. So really looking forward to that
Generally, I am enjoying this. But looking forward to results-based version to try.
When someone mentions Zwift and then qualifies something with “Long Term” we know we are talking , end of time itself long
For me , whilst I can see its possible with more analytics and ML being applied to maybe smooth out some rough edges of a performance predictive categorisation it will never quiet down the baying masses for good reason. If your body metrics is most aligned to the current version of the categorisation system you are going to (stastically) win races , and continue to win races with no consequences and those who were 2nd , 10th , 50th etc will continue to be statistically in those places . What needs to be represented in all this is consequences from winning . OK straight to the chase . Results categorisation . Its also simple to calculatte and arbitrate and understand . So why Long Term ... dont get it
Sure but I think the point is still valid - we could start with a rankings system as the ‘master’ and improve it with physiological metrics for outliers / new racers - that could be done outside of the client too. That being said I do think it would be difficult to really test/implement in an isolated race series like the Classics, A rankings system really comes in to its own when you can jump in to any race / race format (as long as it is set up as a rankings race).
My guess is the most complex part is moving away from categories, as this naturally produces particular problems even in a rankings system. I think Martin hinted at it here:
2 points I take from this (may be wrong):
Once you have a rankings system, how do you determine pens / classes / categories etc? This is a significant challenge without dynamic pens, when you don’t know the field until just before the race start.
The current testing and solution iteration is finding a method to class riders who do not yet have a robust ranking for whatever reason. TrueSkill seems to cover this automatically to some extent, but in combination with some physiological metrics could be even more powerful.
I do trust WTRL to get this right - they are clever chaps and are responsible for the best racing Zwift has ever had in the form of the ZRL. I just hope Zwift alongside them see the value in bringing whatever system is finalised in to the core client for all race types.
I’m pretty generally a bad racer. My power numbers indicate I should be better than I am. For the first two races, I was placed in C3 - I had a dropout in Race 1, but Race 2 I was competitive (9th overall, and some FTL/FAL points). Probably my best non TTT finish in any “real” racing event.
For the crit, I got downgraded to C4. I looked at the competition for the race, and I see riders that are better than me that also got downgraded. Clearly a tweak to the algorithm (or the algorithm spitting out different results specific to the course/style of race). While I don’t like the idea of a downgrade, I am excited to see how this adjustment pans out. I push myself so much harder when a race is competitive, and so many of the races I get placed into by simple A/B/C/D have me out of contention almost from the get-go.
So I’m back in C4 for tomorrow’s West Ozzie Crit City event. As a 66kg old guy this is going to be a challenge. I’m anticipating being on the rivet for the entire race since I’m at the bottom end of the field in 20 minute power. The crazy thing is the top rider is an A cat rider with 4.5w/kg 20 minute and the next two guys aren’t much behind. All three of them put down serious 15 second power AND are big boys ranging from 78kg to 89kg. Doubledraft would’ve been nice…
My question is this: How did these guys get into C4? The rest of the field are more aligned with what I would expect for C4. The three dudes should probably be in C3 or even C2 IMHO…
Been in C3 for a test race and both classic races so far and will be again in tonight’s race. Finishing almost right at the back in every ride but refusing to give up, just treating it as a time trial and keep on pushing. For the next race 13 out of 53 are A riders, the rest are B riders. One rider has 4.74w/kg @ 78kg for 20 mins and a race ranking of 127.
I like that something different is being tried with the categories. Will be interesting when option 3 (race results, etc) get thrown into the mix later on.
At least sometimes my race ranking improves even though I finish so far down.
What is it are we doing now … is this community or zwift ? The community are feeding back but it is zwift that are implementing everything , especially since Zwiftpower was taken from the community and became an official organ of zwift itself ?
I am sort of questioning why we have to have this long term/short term scenario . We are putting all this effort into improving something that will never really be 100% satisfactory and not just go straight to the simpler solution , whether that is as a community effort … or for zwift to do it . Is all this short term thinking actually creating this long term delivery of the solution in effect.
Sometimes you just have to commit to change and doing something and not invest so much time and energy on maintaining the status quo , just put it on fix on fail and lets get results categorising done .
For years we have been asking for Zwift to implement something, now the community is stepping up just like the CVR world cup and Zwift power. Racing was started by the community. Let’s hope that WTRL can get something that Zwift can build on.
The start list for eastern Australia C4 looks grim. Everyone within the 3.0 to 4.2 wkg range for 20 minute power. One guy has 4.1w/kg at 327w, finished 6th and 7th in the first two races in C3 and is now in C4. Another finished 2nd in C3 last week and is C4 this week!
This is not going to end well for me. Taking bets on how many laps I can last.
Right now, Zwift aren’t implementing anything. It’s WTRL who are doing the work, and that makes sense to me in that they’ve got a lot of experience in sifting through the performance data.
ZP is so badly broken that WTRL aren’t even using it, and they’re getting the data straight from Zwift itself.
Ok so there are still auto-cat races going on, but it’s no longer possible to subscribe to them? Am I right? Is there a place where I can find all info centralised instead of scattered out over a 500 post thread?
It will be interesting to see if the .fit file for the races will be saved properly. What with the latest update introducing another Zwift bug that is causing .fit files to be lost.
Is it limited to certain platforms, that issue? I’ve seen the thread rumbling on but I’ve not had any problems since the update personally (Windows 10).
Its WTRL in partnership with Zwift here BTW . Do we think this is a community project. WTRL have in several replies suggested they are not partial to discussing with us what is being implemented as if it was.
I am not knocking anything , just making sure we are all clear on what this is.
Its perfectly good for zwift and WTRL to work to offer something , it will either meet customer approval or it wont and we will either use it or we wont . Our involvement however is purely as customers we either think its worth paying for or we don’t .
Zwift cant hide behind this and pretend they are working on something else long term , this is what they are working on . Make it work for your paying customers
The beauty of it is that it doesn’t really matter much too much exactly how it’s arranged because such a system is self-correcting. If you are getting badly beaten all the time, you drop to easier races. If you win too regularly, you’ll move up to stiffer opposition. Rather than a load of riders getting stuck at the bottom of a category where they cannot compete while others win time after time (as currently happens), literally every rider is guaranteed to have some races they can score points in, some of the time (not all of the time). It gives a varied racing experience to everyone (apart from the very strongest/weakest handful of riders who will always win/lose against everyone else), rather than repeating the same race against the same opponents week after week after week.
