i assume he is talking about people who save their activities as private, which sends a “green bolt” activity to zwiftpower with a sample rate of between 15-30s or so instead of the more precise ~1s
Why is private vs non-private determining the sampling rate of an activity anyway?
no idea. the last bit of coding i did was copy and pasting some premade HTML onto my myspace page back when i was like 12.
2 Likes
Actually, wondering maybe if there’s no sampling at all, and power from a private activity is just being roughly estimated from time/distance travelled for the rider, like Strava would do if you didn’t actually have a power meter.
The difference seems to be the availability of a fit file. Both public and private riders have data sampled live from Zwift, the difference is public riders then have a fit file generated that is then analysed for the moment by moment data. Maybe its something to do with the storage of that data file? That the rider owns that data in GDPR terms? I don’t know for sure.
Regardless, it’s the same problem for ZwiftRacing.app as it is for Zwift Power.
Amused to see that there’s a very clear trend here. Tiny Races improve my ZR ranking. ZRL improves my ZR ranking. WICR reduces it (despite having some results I’m really pleased with - probably just that I’m less good at long races than it expects me to be). But overall it’s not moving by a large amount. Which, I suppose means it’s “about right”.
2 Likes
Indeed. The default display is going to change a bit soon - under 10 point swing either way is basically… no change.
A random thing but a premise under this ranking is that it assumes everyone wants to improve it. People that race multiple times a week don’t expect to perform well at all. System kinda misses that main part for most is training and not performing at their best all times. This causes ranking to be off since people don’t race to optimize their ranking (I don’t at most races) and see significant drops. This is further exacerbated as choosing a race with ranks is not a specific choice, rather it’s just calculated using everything available without selection.
This is different to chess or most online games. Performance here is driven by fitness and we can’t be fit all the time. While in chess I will beat the same players time and time again and the same in FPS games. In a game of fitness, if I just did a massive effort the day before or the same day I can’t perform well again. I still want exercise and I like races to that end but it impacts my rank. When races get sorted based on this it causes quite some noise and randomness.
To that end I like the ZP method more to only look at best performances. Such a base as starting point would improve it in my opinion. Despite them having other problems that as a base seems logical to me since we have to accept that it’s a training platform to most.
1 Like
Do we? You may be right, but then again you may not. I have no idea.
This topic also fwiw has applicability to CE as well. That is, CP established perhaps based on a day/race when when one was rested and closer to peak, and then the same racer continues on and does 4-5 races/week and wonders why they were put in the pen they’re in because of some ‘flukey’ past performance that put them over the top.
1 Like
I don’t see that as a problem. The inverse here, where people will in essence be rated lower, as a problem though.
You are able to perform at that level. Just not all the time, then you won’t do as good as past performance indicates but that’s fine… You will when you have created proper conditions for yourself again.
I think you are applying the self to the mass and that creates issues.
Your rank is accurate, as you perform poorly in lots of races. This is due to the fact you race a lot and for fitness. The intentions are not really relevant though, as the rank is accurate. In any given race I have X chance of beating you. X would be incorrect if it only took your best results in to account.
The ZP system ranks riders based on their capability (how good are they when they are at their best?). It doesn’t do a particularly good job of that for the 90% that don’t race regularly, but ignoring that for now…
Which do you think is the better for organising races based on ranking, and why?
1 Like
Can you point to an example please? So far that’s not the feedback we have had at all, and there haven’t been many races.
Indeed ignore the data issue (your new rankings also apply just a patch for that, nothing is perfect).
I believe in a game of fitness one should be ranked on best performances. Rank should be my best me relative to others at their peak. When we rank athletes we look at them at their peak. We don’t rank Vingegaard based on some preparation race like the Ardèche classic but based on his Tour when looking at recent performances (and I think window of analyzing should be much longer than is custom in Zwift… 60 days, new categorization, 90 days ZP and x days, your ranks, are far to short).
Can expand on this but yes I think it should be based on best performances. I think it will even help people taking ranking more serious and prepare for it instead of it now just being a result using everything. A result which could actually encourage people to race less rather than more. While I’d rather see people race more.
What I’ve been noticing in my use (as someone that doesn’t race that much), is the initial seeding is going to more or less dicatate my rank. The rank ranges are 200 point ranges, and on average my score is not changing that much whether I get completely dropped multiple times, and end up in last place, or if I make it to the end of a race and simply get wrecked on a sprint, to end up middle of the pack (which are my best races - ha).
As an example, the race where I got essentially last place (over 4 mins behind 1st place - was second last, but the person below me entered the wrong category - everyone else dropped out), I only lost 3.91 ranking (partly due to this race happening before the new seeding algo, so with the new seeding I would have been in a lower cat), but assuming that one rider below me didn’t finish I couldn’t have done any worse. So, it would take a lot of those races to drop me into a more reasonable category.
The biggest swing I have out of the 5 races I’ve done in the last 2 months has been 34 points positive, in that one I was 13th out of 33 (less than 1 second behind 1st place) riders which is probably the best I will ever do in any race given I have no sprint and will be back of however many folks make it to the final sprint.
On average my point swings have been ±15, so to go up a 200 point category if I was consistently getting ± 15 would take me over 13 races. The reason I only have 5 total “races” in the last 2 months is the Tour De Makuri group rides don’t count as races, but I treat them as such, there’s always these style of events going on (Tour De Zwift, Tour De Wattopia etc.) so I would guess that for me to get 13 formal races in a seeding session would be unlikely.
All that to say, for folks like me that try to get about 1 race a week, some of those not counting as formal races (such as tour de Makuri, Tour De Zwift, Tour De Wattopia, etc.), the initial seeding is basically going to be my category for that session give or take one 200 point category step. So in that sense it’s basically a power-based categorization for me (compound score, or CP curve, etc.).
2 Likes
i mentioned it in another thread because i’m someone who races nearly daily, sometimes multiple times in a row and i’m not the only one. the short version is that i think a ranking system serves a purpose best if it was a tool race organisers could use to sort pens in their series. i don’t think it would work very well as the default system, CE/ZP actually work well enough for the average daily “i don’t give a ■■■■ about this race” training race
edit: forgot to add, even games like chess and FPS games which have no physical cost have unranked public lobbies. gotta have em
2 Likes
I kind of agree with this point, but I don’t know if it’s because current pens are set by a completely different criteria which is fixed, so you will always tend to score about the same for each race?
With the static boundaries and zwift racing being very same/same it’s kind of understandable - those at the top end, continue to score points and rise, those in the middle will remain as is…
Indeed, maybe the swings need to be more aggressive. That’s an easy thing to tweak!
2 Likes
I don’t disagree with this approach, other than to say that it is a fundamentally different one, and comes with a different set of issues. One of those being that if you rank by best results, riders that haven’t exhibited best results, cannot have an accurate rank. This is a big problem for new riders, returning riders, lower ability riders, etc. It is more of an achievement / reward system than this one, which tries to rank riders in a dynamic and accurate way, with movement to some extent every race, rather than a system that tries to sort all riders by capability.
I think CE/ZP are not working good enough. I can talk about my category (C), and there are so many top racers that should be un B and would be much more fair for the rest of C racers. With better categorized racers, the categories should much more fair.
After more than 700 races, i only consider racer ranking as an option. If you want to beat a 330w FTP / 100kg racer in C as a 75kg racer, you go directly to B. Ranking will categorize all best D, C, B racers in the next CAT and would make much more easy the competition for normal and low racers in their category. Please, give ranking an oportunity. Everything else have failed until now.
2 Likes