Note: for the lastest news on Zwift Racing Score, see our post What’s happening with Zwift Racing Score?
Zwift began testing their new Zwift Racing Score (ZRS) metric in early July, with the goal of rolling out results-based race categorization heading into the 2024/25 indoor racing season.
Since that test launch, Zwift’s game team has been looking closely at ZRS results and Zwifter feedback. Recently they announced plans to launch ZRS in October, while also sharing details of the latest tweaks made to the ZRS system…
Floor Score Change Indicators
Race results now include more transparency about hitting or establishing a minimum/floor score. We’ve received a lot of questions and confusion around why the resulting score from the most recent race result does not match the current score listed in the profile page. Race results… have been updated to accurately show when a racer encounters their floor with new indicators:
Zwift
Here’s what these new arrows look like, and what they indicate:
Your Racing Score increased after this race.
Your Racing Score decreased after this race.
You’ve set a new, higher, floor value for your Racing Score after this race.
Your Racing Score has decreased to your floor value after this race.Why the change?
Your ZRS can change for three reasons:
- Your power metrics change, which increases or decreases your “seed score”
- You finish a race and thus earn a fresh ZRS result based on your finishing position
- You do nothing, and your score decays slowly over time
If you hit a power PR during your race which boosts your seed score, this may impact your overall racing score independent of your actual race placing. With the new arrow indicator, you’ll know if your score is being influenced by a new, higher floor value.
Conversely, you may get a low race result based on your finishing position, and this may cause your overall score to drop all the way to its lowest possible value. In this case, the new arrow indicator will tell you you’ve hit your floor (which is now 15% lower than your seed score, see below).
Increased Score Volatility
We have begun slowly increasing score volatility, or how much a score can change from the result of a single race. It was useful in early days of testing to ensure scores didn’t vary too wildly from race to race, but the data (and your feedback) is showing we can release the handbrake a bit. This change won’t be immediately apparent, but as racing continues, we will start to see an increase in how quickly scores move.
Zwift
Why the change?
Score volatility is a challenging setting to get right when it comes to race scoring. You want scores to change quickly enough that riders feel a since of progress and movement when warranted, while also being quickly placed in the correct pace group. But you also don’t want a single result to affect score so dramatically that it places a rider in an unrealistic category.
Zwift has been testing the scoring system at a lower volatility setting, but they’re finding that some riders aren’t being moved up to their proper category quickly enough. Increasing volatility should help.
New Score Floor
Starting today, your score can drop 15% lower than where you were seeded. Before this, your initial seed also served as your floor or minimum score. Our original seed algorithm tried to undershoot the prediction slightly to account for this, but we’re seeing many cases where people finish in the bottom half of the field consistently and their scores are not allowed to drop any further. This creates a poor racing experience for those racers since they are never able to adjust to their ideal category. Now that the floor is slightly lower than seed, we hope that this provides the flexibility required to ensure the model works for everyone.
Zwift
Why the change?
Simply put, some Zwifters should be moved down a category, but the system wasn’t allowing for that because it kept them pegged at their power-based seed score. This change reduces that lowest possible score by 15%, which will allow some riders to drop further, hopefully getting them into a lower category where they can feel more competitive.
Additional Changes
Zwift says two more changes will be coming in the next couple of weeks:
- A more accurate seeding and floor formula that increases the threshold of what we consider a “quality power curve”: this is a bit cryptic, but Zwift tells me they’re looking for ways to use cleaner data for a more consistent seed score.
- Support for filtering general classification standings on Zwift Power by racing score ranges: showing more ZRS data on ZwiftPower would be welcomed by all, certainly.
Join a Scored Race!
I’m not sure how Zwift will be rolling our ZRS in October. Which events will used ZRS for categories? And will only those events affect your score (as it is now), or could other races that use other categorization methods also impact your score?
We’ll find out soon enough. Until then, try some races using Zwift Racing Score! Here’s a list of upcoming ZRS test races: zwift.com/events/tag/zwiftlabs
Questions or Comments?
Have you been participating in scored races? If so, how has the experience been for you? Post below. You may also chime in on the (massive and growing) Zwift.com forum thread!
It will take many iterations for them to get it right, that’s for sure. The influx of unranked riders returning from outdoor is also going to make it less than optimal for the october release. However, the new system will definitely be an improvement just because it is able to scale properly.
Imo they should get rid of the manual category selection and auto assign each participant while also creating the categories (bands) dynamically based on the participants. The higher the number of participants, the higher the amount of bands. Kinda what they’re already trying to achieve with the high/low races, just dynamically.
30secs and 10mins alone don’t show the true ability of a racer… Who knew?!?>!
Based on many comments/complaints I’ve seen with the initial iteration, I would hope a “quality power curve” would mean deemphasizing the 30 sec power number a good bit in the algorithm. It will be interesting to see how it evolves.
I suppose everyone has the own experience, but my experience suggests that short term power is a big deciding factor in alot of races, especially events with segments like ZRL.
Short term power is obviously a big deciding factor in many races, and if this was another power based system it really should be included.
But in a results based system, the seed score should just be getting people in the right ballpark and let the race results do the fine tuning. Longer interval power is better suited to that purpose, but not everyone will have done a maximum effort over a long interval and most would rarely do so in a race.
The 5 minute time range or zMAP, and a weight normalisation with a bit more to it than simple w/kg, is likely the best solution to getting racers in a field that is close enough to let the results based changes do their thing.
I think that makes a lot of sense.
That is definitely true, certainly important to use it. Seems like as it is now it skews some riders, though. And no system is perfect. I imagine it will keep improving and data is generated and feedback is received.
It adds another layer of complication, but ultimately the courses will have to be categorized (climb stage, sprint stage, etc) and a rider’s ranking should be specific to each course. The current system is like trying to answer the question, “who is better, Cavendish or Kuss?”. There is no way you could assign those two riders a single “racing score” and with any accuracy predict their results on a randomized course. Course profile is a huge variable that is completely ignored in the current system.
Isn’t that the point though? zFTP-based categories tended to put high wkg guys in a higher cat. Beefier sprinters would lower wkg would end up in a lower cat where they’d just win race after race without ever getting promoted, and lower wkg riders without a strong sprint didn’t have much of a chance. By putting Kuss and Cav into the same cat (eg high race score) they’ll both have a chance, just in different kinds of courses. The score should be ignoring course profile; that’s up to the rider to choose.
But as I understand it, the purpose of Race Score is to predict likely race results and group riders accordingly.
It is never going to be particularly accurate at predicting race results and grouping riders if it ignores the course profile in making those predictions.
Think of how you might rank racers in one of Eric’s Tiny Races versus how you might rank them in a race up the Alpe. There is no way the current system can deal with that difference.
Exactly. If you keep winning C grade races, no bones about it you should go to B grade. You don’t have an entitlement to win in lower grades. If after a few races in B you keep losing you get bumped back down. Simple. Alot of the whinging about a results based system seems to be coming from sandbaggers who want to keep winning lower grades.
We are not talking about sandbaggers here; they are regular racers whose endurance levels just do not match B category. What you are suggesting is to penalize them for winning in C by sending them to B. I can guarantee that 95% of them will eventually return to C. But until then, they will not be racing or even using Zwift, which might result in them canceling their subscriptions for 90 days. Smart business choice, Zwift!
C grade is by far the worst grade of riders for coasting/sandbagging, when half the field can easily keep pace and have the same power output as B’s ( over a shorter time ). having raced back and forth in both cats for years now – i see the guys regularly in C who easily could and should be in B – but never are. even their race score is weird. ive checked race score of some of the people who i know can easily beat me, and yet im way above them, and their scores are unusually low – its not me – its them! for some odd reason, the coasters in C are not being caught. you can say, why dont you attack when the entire C- field starts at 5wpk, drops to 2 wpk after a few minutes and the ONLY time they go back up to 5wpk- ot even 6-7 in a lot of cases is at a hill or the sprint. you cant get away coz there are about a dozen or more riders doing this, which as a group easily close any gaps – now lets resume the 2 wpk casual ride until the end.
you go to B however, its kind of the same, but they coast much higher @ 3, they dont lift any harder, just a bit longer. these C riders who are doing this dont want to ride B because they want the easier ride – and then try to justify a B-grade quality sprint as their good work out.
maybe race score should take into account 15 second power also. if you have a high 15sec power in C, you are one of the front guys. if you dont, the only way you can win is to grind off a SMALL field ~ and very few in C can do this (if you can, you SHOULD be in B), and if there’s more than 10 chasing you, forget it.
I am a sub 2.0 D rider, I say this to understand where I’m coming. After doing some races with this new system my life is slightly worse when 10 categories are used, and clearly worse when 5 categories. The reason is that even in 10 categories I am placed in the same category than C paced riders and with 5 categories I have competed against people doing double W/Kg than me. I understand that D riders are few, so we are less a priority than B and C riders. B and C paced riders are the most abundant in Zwift, and as a result of that the new system is going to tend to adjust better to this riders because they have more data, is just something completely normal.
I don’t know if they considered include in some way the actual paced categories with the new ranking.
if you are in the race score of above 300, then you will encounter a LOT of C riders.
@Eric Schlange have you considerd doing and an analysis of how the actual A-D categories get distributed in the new system. Maybe is too soon to do that. I’m asking because I have the impression that as B and C categories had the largers numbers they are know distributed into more categories of the new system.
The entire ZRS system is a complete joke. When you pick random numbers like 30 seconds and 10 minutes, you are basically giving everyone a cheat sheet on how to adjust their power numbers for a lower seed score. In the next few weeks, we are going to witness a crazy drop in sprint powers among Zwift racers. Those with great endurance will start showing up in lower categories and winning races just by cruising in their sweet spot zones. If you are hoping that ZRS will bump them up to a higher category, I hate to break it to you, but that is not happening. The score volatility is just too minimal for any quick advancement. And people can strategically pick important races to win and rubbish ones to lose, keeping their scores under control. I can predict an uproar in October.
It’s odd. I raced in one of the Labs races last night and won the race (field of 19), yet my racing score went down by a point. I don’t mind going down, but it seemed to ignore the longer attacks I was doing.
ive come last, and increased. its weird.
Racing score is a good system, but we need to transition into more mass start events, especially at times where participation is generally low. That will solve a lot of problems for the highest category of racers.
Yup, I ride Zwift between 5am and 7am in US West Coast time. If I want to ride a race, and I ONLY ride ZRS races ‘cuz I get shelled in CE races, I have a choice between a Crit race at 4:40 and a Crit race at 6:40. So I’m going to do the same F****** Crit race every F****** day? The only variation on that is the Tiny Races on Saturday and they go off at 8am but it’s the weekend, right? I have no idea who is responsible for scheduling these events but they probably need to step back for a moment and see if they can serve the Zwift racer better. IMHO, of course…
Since you race in a wrong ZRS category it is just good your choice is limited. Damage control. 😂
I have not the pleasure of understanding your comment. I race in the ZRS category Zwift puts me in. Right now my ZRS is 564 down from 630 a coupla weeks ago. I don’t sandbag and when I’m racing I’m going as hard as I can every single time. If Zwift indicates Cat B is where I’m supposed to race, I go there. I don’t win very often but I’m at least in the mix racing with everybody else. I just wish there were more ZRS races.
If you have got a 4.67 W/kg endurance, you definitely do not fit in the 560+ category. That is more like 650+ endurance. You are basically racing with folks who have 3.0-3.4 W/kg. I get that you are just playing by the rules, but honestly, it is not really amusing. This is a ZRS issue for sure.
You are certainly entitled to your opinion about this, Jessica. I know, speaking from first hand experience that I do not win every ZRS race I enter. I don’t ride away from the pack from the pen and I don’t pull off heroic breakaways in longer races. I am in the mix and contribute to animating the action but have not been able to ride Rog or Pog and play with the peloton. Take a look at my ZP profile, and you’ll see my HR is within about 8 beats of my max tested rate during these races. I literally cannot go harder than I’m going.
When I race in the Cat A CE races, I’m dropped within the first 5 minutes especially if there is any kind of a punchy hill. I cannot hang with real full power Cat A riders because I do not have the short power to keep up. This is why, for me, ZRS has made racing on Zwift something I’m interested in doing again.
Also, I’ve noticed in the ZRS races I’ve done, the peloton stays together more than it does in CE races. Literally, 80-90-% of the riders are bunched until the finish sprint lead up. Longer hills tend to break it up more but when we’re racing in short punchy races you’ve got a sizable pack and that’s not seen in CE races very often.
These are all things I’ve experienced first hand in Zwift racing. I do not know why my ZRS went from 630 two weeks ago to 564. I’m actually completely mystified because I’m not going any slower. But I am not a 650+ rider. Probably never will be.
I’ve just done a few ZRS events and the first notable thing is that the seed score as with the zrCS score is that it dramatically favours low weight racers. These low weight riders start off with a score far below their ability level. The last race I did a guy who average 4.3 wkd absolutely walked away with it. How was he ever seeded that low? His race wkg was a typical effort for him. This type of seeding will blight ZRS for ever if not sorted as there will always be a new or returning light rider causing havoc. It’s the same issue as seen in the current CE system with 5wkg light women in C cat. It’s now just been updated in ZRS to include light males too.
Or was the old system just more weighted towards heavier riders? 2 sides to every coin. Trying to get the balance is key
I’m a ZCE (Zwift Cat Enforcement) A rider because of my long game but no short game. I have a 4.67w/kg FTP and a 7.7 15 sec and a 5.6 1 min. So I get completely shelled in any ZCE A races. In ZRS, I started at 550, got up to 630 last week and now I’m back down to 574; I guess that’s the new volatility.
I’ve won four ZRS races out of 27 with a lot of mid to rear pack finishes. The difference for me is I’m in every one of those races. I may not win but I have been in pack animating the field. Every single one of them. For me, ZRS has made Zwift racing fun again. I will not race another ZCE cat A race again. There’s no point when I can race in an event where I’m part of it instead finishing Lantern Rouge…
Check out my ZP profile: https://zwiftpower.com/profile.php?z=1630774
I was enthusiastic about the introduction of a rankings-based system, but for me it’s being spoiled by the ‘seed score’ which has little relevance to it, and in my case, stamps all over it. I finish near the back in all my ZRS-scored races, but my ZRS score stays the same, because I’m at my ‘floor’. So, I’m destined to keep on finishing near the back – being dropped early and coming home with other stragglers/tankers, or solo-riding. I’ve been racing every week for 18 months, so don’t really see why I need a seed score anyway, but my ZRS is entirely based on it right now, which seems entirely inappropriate to me. So what if I produced a decent 30 sec sprint some time (and I’m not ‘a sprinter’) when I was out free-riding and decided to hammer it for 30 secs – what’s that got to do with racing? If I do manage to stay with the bunch until the end of a race (it will have to have been a flat course), I’m totally f***ed with the effort of doing so and can’t sprint when needed anyway.
And another thing, I think (and I’m talking generally now), a rider’s ZRS should go up ‘much more quickly’ than it comes down. A good finish is more solid evidence of the presence of ability than a bad finish is of lack of it. And I’m thinking here of riders who do ‘one race on, one race off’ to keep their ZRS constant – winning, then tanking, in alternate races. It’s too easy for them to stay in a category this way, particularly when ZRS isn’t checked at the time of race-start. I’m already seeing it, and there’ll be more of it to come as ZRS-scored races become more widely adopted. With ‘quick rise, slow decline’, at least they’ll have to tank multiple times in succession to get their score back down to where they want it.
It still seems that zwiftracingapp is a far better ranking system than the zwift racing score. Last night’s pinarello round 2 whatever, zwiftracingapp was pretty much bang on on predictions vs actual finishes.
Yet with zwift racing score, the guy who was pushing the front of the race and has won or podiumed 14/20 of his last races has a lower ZRS (425 vs 483) than me (5 podiums out of 20 races – 1 win was a race of two people).
For reference zwiftracing app has him at 1211 vs 1119 for me.
It would be great if there were more racing score events. I can’t make most of the odd times they are offered.
It gets old racing in regular B with a the # of As that are in B.
There are quite a few Bs in regular C events as well.
Those are easy to spot in zwiftracingapp.
For racing score events, A riders shouldn’t be in “C” ZRS events and Bs shouldn’t be in “D” ZRS events, but it happens all the time because certain riders have a low racing score.
Racing score does need further tweaking as it does appear to be easily manipulated. Effort below a certain threshold shouldn’t decrease your score for example.
Not just a simple tweak. ZRS needs a fundamental change, a total rewrite.
I find the races way more entertaining but it looks like there is still some work.
I steadily increased my RS to 417 in 350-400-450RS events. Now I finish 23/90 in a World Series 350-520RS event, best power in months, and the results show RS 408 with an arrow indicating my RS increased (not my floor) whilst everyone above me has a RS of 450+…
I should expect a RS and floor increase here to be fair. As Zwift itself states that racing against larger fields is beneficial for increasing your score.