It was big news when Zwift began testing their new Zwift Racing Score (ZRS) metric in early July, because that meant results-based race categorization could be coming soon.
Now, a month into the tests, statistics and racer opinions have emerged which are in turn driving changes to the ZRS system as Zwift works to make it ready for a larger release.
Let’s take a look at some of the numbers, opinions, and changes made or planned. But first…
Take Our Poll!
I’m curious: if you’ve tried one or more scored races, what do you think of Zwift Racing Score? Here’s a one-question poll where you can rate your experience:
[ays_poll id=”2″]
Some Stats
Zwift staff recently shared a forum post that includes some ZRS stats and plans (see it here).
As of July 30, over 12,750 unique racers had joined over 25,000 scored races. Most riders have finished just one scored race, which makes some sense, since there aren’t a lot of scored races available:
The distribution of racing scores is as follows:
Summarizing Feedback
There’s a lot of feedback flowing in from the ZRS test events, with riders posting comments in the Zwift forums and the Zwift Labs club.
I certainly haven’t read all the feedback… but I’ve read a lot of it. Here are some of the themes I’m noticing:
- Race groups seem more diverse. That is, you’ve got riders with strong 30s power and relatively low FTPs mixed in with riders who have higher FTP but a poor sprint. This should result in more interesting races as the high FTP riders put in hard, longer attacks to try to drop sprinters before the line.
- Heavier riders and/or those with big 30s power feel like they’re seeded too high. Some riders with a strong sprint (in terms of pure watts) are complaining that they are getting dropped quickly in their races. This seems to be more than a “you’re at the bottom of your category” issue, and more of a “your seed score is higher than it should be” issue. But it could also be a “you’re having to work for the result instead of just sitting in then sprinting” issue.
- Seed score increases are confusing some riders. Your seed score is based on your 30s and 10-minute power, and you can set a new PR for either of these in any Zwift activity. It’s very possible to set a new PR in a race and have your ZRS change based on a new seed score as well as your race result. This could be a bit confusing, and Zwift may need to make some UI changes to improve the experience.
- (Possibly) unrealistic expectations: I’ve seen various comments from riders whose expectations don’t seem realistic. A few examples:
- “My ZRS band is harder than my old category enforcement category, and this is wrong.” It’s two completely different categorization systems, and nothing says that you should be at the same place in the pecking order in both.
- “My score isn’t changing enough.” If the ZRS system is accurate, your score shouldn’t be changing drastically between events. Your ZRS is supposed to predict race performance, and if it does that well, you should finish at or near your current ZRS.
- “I got dropped. The system is broken.” Lots of riders get dropped. You may have been at the top of your category before, and now you’re at the bottom. Some riders with strong sprints have been sitting in and sandbagging the lower categories and are now being forced to race against stronger riders. ZRS wasn’t meant to make racing easier: it was meant to make it fairer.
ZRS Changes
Zwift’s devs have already made some changes to ZRS since its initial rollout, including:
- Slowing the rate at which your ZRS decays
- Fixing some UI bugs
- Showing your updated score when you return to the game homescreen after a race
- Changing ZRS-based bands in ZRacing events to use different score ranges
Additional changes are planned for August:
- New indicator icons in race results to show if you hit your floor value or had a major jump based on setting a new 30s to 600s best
- “We will also be fixing an issue that prevented score from moving as quickly as they should when a new 90-day power best/floor is established”
- Fixing web results so up/down arrows display correctly
Join a Scored Race!
If you haven’t tried a race using Zwift Racing Score, give it a try! Here’s a list of upcoming ZRS test races: zwift.com/events/tag/zwiftlabs
Questions Or Comments?
Questions about how ZRS works? Comments about it? Post below!
The biggest issue I see with the seeding system is many people do not have their max efforts recorded in those individual 30s and 10 min durations. Which causes some really good riders to be sorted in way lower “category” than they should be and for these riders, the rate of growth isn’t fast enough to get to their actual level in a reasonable timeframe.
Other than that, I love the new system and thing it will improve a lot of problems once it goes live.
Would help massively if Zwift would actually keep the ZwiftPower site (or a replacement if they’re done with it!) up to date, currently on event results pages there’s no ZRS column for EventScore/Before/After/Gain/Loss etc there’s just the old “Race Ranking” 600-1 score, and when looking to sign up for an event there’s no column at all to indicate to you what ZRS other riders have, nor does it show you 30s or 10mins power for those riders, so it’s hard to try and pick races where you will be able to actually feel like you can genuinely compete without getting massively dropped “while trying”, and I think that makes people feel like they don’t want to bother with those types of events. Lack of clarity leading to lack of enthusiasm over racing and then lack of involvement, which sets a worrying cycle going forwards
No. It’s just like a Cat4/5 race in the real world. You never know who shows up and in what condition. So you want to cherry pick the race where you’ll do well? Race a few times and you’ll figure out how to race. If you can’t sprint take a flier and drop the guys that only sprint.
I get what you’re saying, but online racing isn’t the same as real life, plus ZRS as a rating system is obviously going to get compared to its predecessor and alternatives, and both of those provided people with information to allow them (as much as anything) to gauge which other registered riders were more Sprint biased and which were more longer distance “Diesel’style riders. When you can’t see/hear other riders next to you online racing needs to give more info to allow.you to gauge your competition, or at least I think so and I know a lot of others do as well.
Is there a way to opt out for those of us who don’t race?
If you don’t race this doesn’t affect you in any way. (Except for the second order effect “if there wasn’t racing on Zwift, they could spend their time doing other things you may like more. But if there wasn’t racing on Zwift, it would be mych less popular and they’d have far less money to spend on anything.)
I just don’t want it on my profile… That’s all… I should be able to opt out of it like Zwift Power.
Hi, Eric. I have a suggestion: change the content of the question to make it clearer that it refers to the previous system adopted by Zwift since there are races based on zwiftracing.app which are better than the ZRS, IMHO.
Done!
The new score system is better than the old category system, but I still think they’re over-complicating this.
Just purchase zwiftracing.app and have that be the ranking system! Done.
I don’t know what that would cost, obviously, but it sure seems like they’re investing a lot of research and development hours to try and recreate a system that is already made, already works, and has very little complaints.
This has been a real head-scratcher for me too.
But I think they realize the mistake of started from scratch a year ago. At least the person that runs Zwiftracing.app is a consultant for this system.
I hope this new scoring system gets there at some point, but it will take time. We need a much larger sample size of courses before knowing how it matches up with Zwiftracing.app. But it’s a good sign that they’re iterating as they go.
Fingers crossed!
That’s what I mean, though. They’re likely paying him as a consultant to help all of their developers that are being paid to recreate it. Just buy the dang thing! Haha.
Not to anger you more…but I was on a group chat where I believe he said he was willing to give it to them for free or next to nothing.
He’s not doing it for money, he just wanted to build something cool and create a better racing experience.
However, we are were we are. I’m staying hopeful that with enough time, testing and iteration this new system gets to where it needs to be.
Have a look at Zwift CEO Eric Min’s racing score and recent results. Does this system put him in an appropriate category? Is his presence and performance in those races fair for the other competitors?
Eric Min (getting a start in cat D and winning solo by 2 minutes by doing 3.9w/kg for the race) really highlights how poorly ZRS is ranking people based on their 30s power and I don’t think Eric Schlange mentions that aspect at all (beyond mentioning the people being ranked highly because of a strong sprint). People with 4w/kg FTP just shouldn’t be getting a start in D category regardless of how bad their sprint is. I think the 30s power number has way too much influence over the seed score and Eric Min’s recent results prove it.
its completely busted. ive got plenty of data, and its bad.
for 1 experiment, i coasted @ 2wpk, finished dead last in C, puddled around – but did 1 sprint which landed me fastest in C for the sprint which gave the race score of 600. way more than everyone else.
i’m the lucky winner of racing with the B’s and A’s for doing 15 Sec power.
interesting to see who voted down and what your race score currently is and what cat you were?
I can help you out here. The reason you are being downvoted is that what you write makes no sense. You entered a race, didn’t work hard and finished dead last. We were with you that far. But then what? Are you talking about what happened in the same race or in some other race? And what happened exactly? Completely incomprehensible. Please try again.
simplified; my score went up for not trying. and its happened TWICE.
What is the point of decay? Riders go away for a couple of months/between ZRL rounds and come back just as fit, why would their score decay? If they’re not as fit then their score will go down eventually including their last seed point. It just seems another way of sandbagging the big series,
Otherwise It’s a massive improvement even if I’m coming last in pretty much every race. I won a lot in the old system, it’s fair enough that I get the hammer.
I had surgery, got complications, spent a week in the hospital, 2 months off my bike, lost 30 W FTP and gained 5 kg this spring. I should not immediately be put back in where I was prior to surgery. That’s why decay is important.
There’s always exceptions, your stats would reset after 90 days and your score would deteriorate. Far more take the summer off Zwift, come back just as strong, often stronger and rip up the lower categories.
Paul, your situation is the exception and not the rule. I was in a similar spot recently. Something as simple as “contact support to wipe your power curve in extenuating circumstances” addresses it. Otherwise there should be no decay, at least for a year.
It will be interesting to see if Zwift wants to use ZRS for ZRL – that would potentially be a big shakeup and where I see some of the biggest benefits for the new scoring system, though there will probably be some rough patches as people will come in with their seed scores that may or may not accurately reflect their Score.
They’ve already said they won’t use it for the next season, but haven’t ruled it out for subsequent seasons this year. It’ll mean a lot of teams will need to change, and the finals format will have to change too.
Once it’s ready I hope they use it, even though I am massively advantaged by the previous system.
Bummer. I was hoping they’d use it this year. That likely means I’ll be focusing on the DIRT series that uses zwiftracing.app.
I did not read all the feedback in the Zwift Labs club, but if these are most of the feedback, I’m very very happy.
It is a very honest way of ranking: you get ranked on results (ok maybe your seeds score as well). This results in that chances are that your finish as often in the top 10% as in the bottom 10%. It may result in some races where you have the lowest ranking in the field. I hope everyone can understand that reality and enjoy the learning in racing with stronger riders.
To be honest: i don’t think this comparison is fair. Use the same amount of racing catgeories in the old system as we have in ZRS races and most people will say it is better. I don’t think the improvement is coming from the new scoring system.
Switching from 4 categories to 5 has had a far bigger influence at the moment than ZRS I think. In B category this has meant a lot of people who would be ranked low in the old A category have dropped down into the new B cat. Same with the new C cat, a lot of low B’s have dropped a cat. For me, as a mid old B rider that has meant I am now a low new B so the racing is harder but that’s not because of the scoring system, it’s because of the redistribution of riders into 5 cats instead of 4. Not surprisingly, the people who have dropped a cat and now find themselves at the top of their new category think it’s a good system.
To be fair I really dont see much improvement overt the old system after a month of testing.
Main problems we had were :
1 – Gaps between the riders in cats were too big. – This hasnt been solved, and it sure looks like it has only gotten worse.
2 – Too few riders in the top cat. – Not solved at all. Cat 1 and Cat 2 are still empty, even at prime time races.
And with the routes becomming more important I see people skipping races where they know they wont do well in. We need more racers, not less.
3 – Too few cats. We need more, to solve a bit of problem 1 but we also need a lot, and than a really lot more people racing than currently. Somewhere between 10 and 15% of all Zwifters race. If there were way more people racing, that cats could get narrowed and people would meet more riders of the same capacities.
4 – Seed system is a mess. You just cant compair 30s and 10min values and tell the riders that they are the same. Its like putting Harrie Lavreijsen and Remco Evenepoel in the same race because one has a sprint and one has endurance and that those values makes them equal.
5 – Sandbaggers – Was a problem and I know it because I was one. Travelling between B and C I know how it feels. But this problem isnt solved and perhaps even gotten worse. Imho there were two kind of sandbaggers. One group who thought 1 was the problem and the gap in the next cat was too big, and a group who just got s used to winning they dont want anythings else.
Both groups will still be here since those problems arent solved. You can still sandbag your way through the cats.
And added to that are now the outside summer riders who can come back and very slowly move through the cats when they return if they dont do stupid 30s and 10 min values. When someone hasnt zwifted in 90 days and takes it easy it will take months before he reaches his limit if he takes it easy but probably will do very well in all of his races. If you just get 8 to 10 points are race it takes a very long time to get from 100 to 600 points. Certainly with the decay we have every day.
You make some very good points. However, it MAY not end up being as bad as you think.
Race organizers can now choose the number of pens (I think up to 10?) but also tailor each pen’s range to their liking. While Zwift currently uses a range of 425-550 for one of its pens, Eric could decide the Tiny Races are better served with a pen of 375-450. Races with varying ranges SHOULD reduce the incentive of hugging the high side of categories (ex. staying just under 3.2 W/KG).
We are also still very early in the testing phase. By increasing the number of races and variety of courses tested, they should have the data necessary to iterate properly.
I do agree that: this is a hard time of the year to get good data, riders often avoid races they won’t do well in, and because of that, scores will often be a better reflection of how good someone is in a specific style of race (flat with sprint finish) as opposed to their score as an all around rider. I’d be in favour of different scores for each style of race, but I’m likely in the minority on that.
Personally, I still think they should have just adopted Zwiftracing.app and iterated on it.
However, as that’s not the case, even just allowing race organizers to tailor the number pens and their ranges will improve a number (but not all) of the concerns you have.
I dont think we have the numbers to make 10 pens for every race.
Thats why I mentioned that first of all we need way more racers.
That’s fair.
I can see 5-7 being optimal for 90-95% of races.
Do you think some of my other thoughts make sense – or am I being too optimistic?
I didn’t realize organizers would be able to choose the amount of pens. That’s a big addition right there.
At the end of the day, people are going to cheat.
I agree with you that zwiftracing.app is the best option (for now), but I do think the new ZRS is better than the old one. And being able to add more pens makes it even better.
More pens might mean fewer people per race. But it also means less people are effected by the few amount of cheaters there are. Plus has anyone else done Club Ladder? 5v5 racing and it might be the most fun series out there. Small pens can be fun.
An alternative is dynamic bands. Simply only sort riders into a number of bands just before the race starts. Base it on the number of riders, with more bands if the total number is big enough. That way you won’t consistently be at a particular end of the scale.
I’m a big fan of the dynamic bands idea, Harmen. Been talking about it for years!
The biggest problem with the old category system was that it was too wide. This lead to people being unwilling to go up one.
This will make that worse, not better. Good sprinters will end up in categories with people with unbelievably better endurance riders making even flat races just a case of hanging on, and anything with a climb pointless.
The best bit about Zwift is the tactics and how hard you need to think to do well.
This will make it worse.
A small skew to push people holding back on the very top of a category, 100% would be great.
Skewing this great is just stupid.
I am a low B / high C on the old system for both metrics.
The new system puts me against A cat riders. (Score of 615)
Essebtially punishing me for racing well. Great…. thanks Zwift!
Big red flag for me in the “number of ZRS races per racer” chart. We heard at the start that the system needs ~5 races to get you to your right place. From the looks of it less than 10% of riders have done that many. So, instead of testing the results-based aspect of this system, we’re just having a bunch of people with similar “seed scores” racing each other once or twice.
When I look at the results for the ZRS races in ZwiftRacing that I’ve done, we still have Golds, Amethysts, and Rubies in the same pen, same as in CE B races. In effect we are just testing a slightly different power-based system. Without more data going into the model we aren’t seeing it getting tuned any better.
Honestly would say we just rip off the plaster and calculate ZRS based on every race result out there, regardless of whether the pens are ZRS limited or not. That will start sorting people into the right pens and let us really know how well this is working.
They would get more people if the new ZRS was included in the Zracing GC competition. Or if it had it’s own GC competition.
I think a 5th Cat for larger series like ZRL would help even things out also & would be more achievable / transparent with the new points system.
The biggest issue is that Races over 750 ar empty all the time! The A races were empty before but now they are even more empty…
This is driving me crazy too. I mainly zwift race to get a good workout when I don’t have time to get outside. Level A. ZRS 743. As a kinda low A, I don’t mind racing against folks who are way better than me. But I do mind not being able to find a race with anybody in it when I have time to get on.
There would be a revolt, but I would love to see it just broken down by quartiles or quintiles. You just sign up for the race, not a category. In the start pens you get sorted into top 20%, next 20%, etc, etc. So all fields are the same size. Folks would go insane because there would be no parity in the top and bottom categories. But at least there would be someone for everyone to race against, even in the smaller races.
this is the biggest issue for me as well. I just can’t find any races with people in them. Very close to dropping Zwift because of it. If there are only a couple people in the A race, combine fields. It really doesn’t seem that hard.
I personally don’t understand alot of the complaints I read about ZRS.
Don’t like the seeding? Don’t worry, it will fix itself up over time. Just get out there and race and let it sort itself out.
Races are harder now? That’s probably because your racing style was helping keep your numbers down. If you genuinely can’t handle the score you got, it will drop by way of your results and you will end up where you belong.
Think it’s unfair? This is exactly how grading is done IRL! Your results determine the field you race in from club level all the way up to Olympics. If you win D grade crits at your local club by a lap, do you think they will let you stay there? Of course not. Ultimately I love reading comments about unfair, these are people most likely riding to maintain their grade below where they likely belong.
That’s how it should work, and once zwift have figured out their final version that probably how it will work. But that’s not how it works now.
There is currently no way to drop your score below the seed score except for waiting up to 90 days for the power record to expire.
And currently whether someone has or hasn’t done a sprint in the last 90 days has a significantly bigger impact on their score than their race results.
Zwift have acknowledged both of these issues. They are real and they are being worked on.
Hey Eric keep up the good work on providing information , especially to me a newbie since Dec 2023 . I find if I race in CAT C and there are lets say 30 people I might be 25 . If I race in the ZRS if there are 10 I might win or pace in top 3 . The reason why is for C I’m older at 67 so I expect to place towards the back . In the ZRS my 232 is on the high end close to 250 so going against new racers who have much lower scores I do well . But in the same race my time is better in C than ZRS but I place better in ZRS . Each has its pros and cons . I get something out of both .
I have raced a fair number of these now and my score seems to be fluctuating up/down around 300 which puts me at a category boundary in the split high/low races.
Profile is of a C rider but capable of holding B on flats as have endurance but low absolute power and NO sprint. B or C, I’m stuffed in the z2 and sprint format…
In ZRC races I do well on where with a long climb or where have opportunities to inflict pain with sustained efforts. Anything with a punch still puts me at risk and anything with a sprint – still stuffed. However Races that go to a sprint I am dead and buried. However Even in races that are not my sort of course, I still feel I can do “something” vs a forgone conclusion.
I think my profile is one that was accurately judged for the initial score. Racing lots and in different types of races that both suit and do not suit me means I am at the right level.
In short – new system working well.
I find that races based on zwiftracing.app ratings to be a lot closer. The couple of Zwift race score races have — given me races with racers from level 3 to level 8 zwiftracing.app levels — and pretty much that’s how people finish. The level 3 types take off on a break away and no one can keep up.. So I’m really liking the app scores better now. I like the Dirt series and ladder races being based on the app scores. Much closer racing.
This poll is a bulls**t. Out of 12500 people, 8000 tried and never returned back. This fact alone speaks volumes. If the experience was really that great, the retention rate would not stand at 36%.
Being on Zwift for 7 months and 67 yrs old , cat c , the bottom or close to it . Under 250 ZRS top or close to it . Over 250 bottom again . But I enjoy training and racing , but like outside Duathlons , its really racing people my age division . And with that always in mind I’m at the top or close to it . So either way for me its the same .
I would suggest the old category system needs to reflect a riders entire power profile including their sprint categories rather than merely FTP. The FTP based system really only categorises one type of rider disproportionately benefitting only those with a kick who can roll around in lower categories than their results support. Those suggesting that a results based system is unfair to sprinters mustn’t race in the real world with clubs. If you keep winning C grade at a certain number of points the club handicapper puts you in to the next grade. This helps keeps the racing interesting and exciting for everyone. For example, its really boring when the race breaks apart and you get stuck in a chasing group with 1 or 2 riders who 5 km in to a 30km race want to sit on for the entire race waiting for the sprint for 8th….
I have a challenge when starting a race I pedal like nobody’s business but get dropped in the first 100 m of the start and have to work super hard to catch up after 7-10kms why is this then I’ve spent or burnt many matches and suffer to the end
“Some riders with a strong sprint (in terms of pure watts) are complaining that they are getting dropped quickly in their races. This seems to be more than a “you’re at the bottom of your category” issue, and more of a “your seed score is higher than it should be” issue. But it could also be a “you’re having to work for the result instead of just sitting in then sprinting” issue.”
I was a Cat C rider with 2.8 w/kg FTP. I podiumed nearly all of my races. As a result, my ZRS has me racing against primarily A and B category riders who are far outside of my level.
Sure, I was at the top of my category and now I am at the bottom but I can’t simply “work for my result.” In my most recent race, I got dropped in the lead-in doing 3.5 w/kg (0.7 w/kg above my FTP). I finished 8 minutes back (in a 43 min race) from 9th place who averaged 4.1 w/kg. For me to crack top 10 in a 17 person race I now need to increase my average watts by 120 or maintain my same power output and lose 29 kg.
Both of those scenarios are unrealistic and is not a matter of me sandbagging or needing to “work harder”.
For some, like myself, this update has ruined the application. I pay $20 / month to race, not ride by myself. If I want to ride by myself, I use Rouvy which has a far more appealing UI for solo rides.