Thanks to all who have joined in on the Racing Score test events and gave us their insights! After four weeks full of exciting races and analyzing your feedback, we’re thrilled to share some updates with you. Your input has been tremendously valuable! We’re excited to continue testing and tuning until the eventual graduation of racing score from Zwift Labs.
To date we’ve had over 12,750 unique racers join racing score events over 25,000 activities.
We’re keeping a close eye on how scores are moving. Here’s a glimpse at the current distribution.
We’re continuing to do a deep analysis of the race result data to inform decision making around how volatile we want scores to be, whether there are biases to under/overseeding in certain populations, and what further improvements we want to make on the data leveraged to seed first time racers.
New and upcoming
Zwift Game will now update the score that displays when returning back to the home screen after a race
August test events will begin to experiment with different subgroup score ranges
Soon, race results in web and Zwift Companion will reflect floor values along with new arrow indicators to make it clear when a racer hits their floor or establishes a new one.
Currently, it feels like 30sec and 10min power seeding scores are dominating race results, as far as Racing Score changes.
It feels like seeding scores are being dominated by 30sec raw Watts power.
Where are racers putting out true max effort 10min power to affect their seeding score?
True aerobic ability appears to be completely ignored right now.
Seeding scores and the pens racers are currently being put in suggest W/Kg isn’t being given anything like equal weighting against pure Watts.
While essentially flat races seem to be giving relatively reasonable races at least in pen E, where heavier racers have a chance to sprint for a top ten or so finish among the lightweight racers in the leading group, routes with increasing amounts of climbing are giving relatively awful races.
If there isn’t going to be a score modifier for flat/ rolling/ hilly/ mountain routes like ZwiftRacingApp has, wouldn’t it be better to initially seed racers on their ability on route profiles near the rolling/hilly boundary?
Given how most participants have only 1 or 2 test event races wouldn’t it be reasonable to also enable a view where ANY race impacts your racing score, so there are more people who actually have a chance of moving up/down based on race performance even if they didn’t make it to a specific test event?
There will still be a lot of folks with only 1 or 2 races etc, but maybe at least a bit more would be showing more movement based on performance.
Because this is a test the only events that would actually use the ZRS during this time would be the test events, but then there would be at least more opportunities to see changes in ZRS based on other events too.
What can we expect to be any different in August events?
I race quite a bit but the ZRS races are not really my go to series so what incentive is there to keep testing other than to help get more data. But with weeks of data i’m not sure what you are looking for when you say you are continuing to do a deep analysis?
ps please please please don’t be testing different subgroup score ranges with the intention to use fixed ranges forever in zwift events.
Id go with that, the riders who really need to be using this platform to test/break it are staying away from it. As you have both said, there is no reason to race them add in the message ZRL wont be using it, what incentive is there?
Some dont want to race harder competition, some like the status quo of collecting their little cups in the CE races.
that’s purely a zwiftpower issue. the joys or having not replaced ZP over the last few years is coming back to bite still.
Edit but i do suspect it’s probably a total pain in the a** to change events to ZRS and set bands with their internal tools so still a lot of work for the events team (probably James)
and yeah just make all of zmonthly events use it if it’s a data quantity issues and you need more data which is what is being suggested.
Some dont see the point in testing when you get categorized from 30s values done in a 1.5w/kg group ride where you sprint when fully rested. And than those 30s seems to be the most important value in the calculation. Staying as far as I can from those test events.
Yep that’s the main concern most riders are still mainly placed on seeding given there is very little volatility in riders scores. I think returning riders after summer can cruise to wins for months and months with the current system.
given the length of it relative to 30s it’s probably not a huge deal if someone is 10 or 20w off their PR over 10 minutes. neither in terms of what category they end up in or in terms of what they can actually do with that during races, since one good effort is gonna yeet them right back up where they belong anyway
mine hasn’t moved since race 1. i guess there is an argument that people winning a lot should be moved up faster irrespective of what numbers they put out, but being able to move up faster means people will probably have to be able to move down faster also. i cant really say from personal experience, since from what i understand the tiny races have some unrelated issues with scoring
Certainly looks that way from here. I had a bunch of very hard races with a few 10min PRs last month and my racing score ticked up a bit. Then I did some hard sprints during an otherwise-easy group ride and my ZRS went up 100.
If you look at how many of the racers have only done 1 or 2 races it somewhat showcases the issue that Zwift has to deal with here. It only takes one or two folks to blow up a race, so if a large portion of racers have only done 1 or 2 races you really do need to be pretty strict with seeding to avoid lower categories being a complete mess.
That said, they can make the gains from winning much bigger to compensate for that. But, when people are seeded with folks all about their same strength it’s hard for an ELO system to really boost the gains a huge amount. If there were seeded events that had all cats start at the same time, then you could see a much bigger ELO boost as you would be able to see people beat folks 2 or 3 categories up if they were not seeded well at all.
I do think regardless of what they do here they should really consider normalizing flexible categories by default to avoid someone sandbagging to be “top B”.
Only zwift will know, but I’d hazard those racers either new or returning probably don’t have that accurate power metrics, so regardless seed score is going to be inaccurate and place them in the wrong category.
Once in the wrong category, there is not enough volatility in the scoring to move them out of the category and into the new one.
Riders can win and not set power pbs when under ranked so they don’t get a new seed score.
The result part of all this should be the primary metric, currently it is entirely seed centric.
Very much agree, but I think I have seen zwift say they plan to have static boundaries in their events but community can move them.
The best Zwift can do is take all of their knowledge of your activity into account when assigning a default pen (ahead of having a good amount of race results). If they know you can do 400W for 10 mins at 65kg because of a workout you did then there’s no reason to let that person into the lowest pen - so the seeding by power is helpful and will catch some folks who would otherwise be throwing themselves in a lower pen.
It’s true that a lot of folks won’t have max efforts and can skate within a lower cat as a result, so if they win hopefully the scores can be boosted to help with that.
So I think Zwift needs to do the best they can to seed people based on what they know about their power regardless, but maybe they can be more aggressive in pushing people up to the next category if they get ‘top ‘x’’ in a race.
I mean, at the end of the day if everything were completely “fair” I guess half the time you’d be in the top half of your field, half the time you’d be in the bottom half (excluding folks at the absolute best and worst racers overall who will always be the best or worse in any cat system :)), and that simply isn’t at all the way it works today, and with the small changes in racing score there will still be people who for the most part are always racing in one category either always top half, or always bottom half.
